RCPSP/RCPSP_S_context.csv

task_name,problem_type,instruction,instance,solution,obj,instance_variant,solution_variant,context_index,input_format,input_index_base
RCPSP,RCPSP,"We’re managing a construction schedule where every task takes a fixed amount of time, uses some limited crews and machines while it’s underway, and a few tasks must wait for others to finish first. The choice is simple in concept: give each task a start time (exactly once), make sure overlapping tasks don’t exceed the available crews or gear, and arrange things so the whole project’s completion — the time the final task ends — happens as soon as possible. The specific jobs, resource limits, and precedence notes follow below.

We have 9 activities, 1 resource types, and resource capacities .
Activity T1 runs for 1 time units, requires 0 while active, and its successors (which must start after it finishes) are T2, T3, T8.
Activity T2 runs for 3 time units, requires 1 while active, and its successors (which must start after it finishes) are T5.
Activity T3 runs for 2 time units, requires 5 while active, and its successors (which must start after it finishes) are T4, T6.
Activity T4 runs for 1 time units, requires 3 while active, and its successors (which must start after it finishes) are T5.
Activity T5 runs for 1 time units, requires 2 while active, and its successors (which must start after it finishes) are none.
Activity T6 runs for 2 time units, requires 6 while active, and its successors (which must start after it finishes) are none.
Activity T7 runs for 4 time units, requires 3 while active, and its successors (which must start after it finishes) are T9.
Activity T8 runs for 5 time units, requires 1 while active, and its successors (which must start after it finishes) are T9.
Activity T9 runs for 1 time units, requires 0 while active, and its successors (which must start after it finishes) are none.
We'll assign exactly one start time per activity so concurrent demands never exceed  and the project finishes as early as possible.

When you send the schedule back, just use this simple JSON shape so everything’s clear and machine-friendly — nothing fancy, just that layout.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that like a little form: ""solution"" is the list of scheduled jobs, and each job record says which task it is and when it starts and ends. Keep it light — the JSON above is just the shape I expect, not the filled-in schedule.

Please make sure to use the exact identifiers from the instance input — do not rename tasks or invent new labels. For example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 2, 7]}, {'duration': 3, 'demands': [1], 'successors': [4]}, {'duration': 2, 'demands': [5], 'successors': [3, 5]}, {'duration': 1, 'demands': [3], 'successors': [4]}, {'duration': 1, 'demands': [2], 'successors': []}, {'duration': 2, 'demands': [6], 'successors': []}, {'duration': 4, 'demands': [3], 'successors': [8]}, {'duration': 5, 'demands': [1], 'successors': [8]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 4, 'duration': 3, 'demands': [1]}, {'task': 2, 'start': 4, 'end': 6, 'duration': 2, 'demands': [5]}, {'task': 3, 'start': 6, 'end': 7, 'duration': 1, 'demands': [3]}, {'task': 4, 'start': 9, 'end': 10, 'duration': 1, 'demands': [2]}, {'task': 5, 'start': 7, 'end': 9, 'duration': 2, 'demands': [6]}, {'task': 6, 'start': 0, 'end': 4, 'duration': 4, 'demands': [3]}, {'task': 7, 'start': 1, 'end': 6, 'duration': 5, 'demands': [1]}, {'task': 8, 'start': 6, 'end': 7, 'duration': 1, 'demands': [0]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T3', 'T8']}, {'task_id': 'T2', 'duration': 3, 'demands': [1], 'successors': ['T5']}, {'task_id': 'T3', 'duration': 2, 'demands': [5], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 1, 'demands': [3], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [2], 'successors': []}, {'task_id': 'T6', 'duration': 2, 'demands': [6], 'successors': []}, {'task_id': 'T7', 'duration': 4, 'demands': [3], 'successors': ['T9']}, {'task_id': 'T8', 'duration': 5, 'demands': [1], 'successors': ['T9']}, {'task_id': 'T9', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 4, 'duration': 3, 'demands': [1]}, {'task': 'T3', 'start': 4, 'end': 6, 'duration': 2, 'demands': [5]}, {'task': 'T4', 'start': 6, 'end': 7, 'duration': 1, 'demands': [3]}, {'task': 'T5', 'start': 9, 'end': 10, 'duration': 1, 'demands': [2]}, {'task': 'T6', 'start': 7, 'end': 9, 'duration': 2, 'demands': [6]}, {'task': 'T7', 'start': 0, 'end': 4, 'duration': 4, 'demands': [3]}, {'task': 'T8', 'start': 1, 'end': 6, 'duration': 5, 'demands': [1]}, {'task': 'T9', 'start': 6, 'end': 7, 'duration': 1, 'demands': [0]}]",1,nl,1
RCPSP,RCPSP,"There’s a rush night and the kitchen needs a timeline: each recipe step has its own runtime, ties up certain tools or staff while it runs, and can only kick off after specific earlier steps are done. The aim is to arrange start times so the entire meal service ends as early as possible — you read that off as the clock time when the last step finishes. Make sure no step is forgotten or duplicated, honor the “finish-before-start” needs, and don’t ever have more simultaneous demand for ovens or cooks than on hand. The detailed list of steps, times and resources follows below.

There are 4 steps, 2 resource types, and the resource capacities are .

| resource_id | capacity |
|---|---|
| A | 4 |
| B | 4 |

| step_id | step_duration | resource_demands | successor_steps |
|---|---|---|---|
| T1 | 2 | 0, 2 | T3, T2, T4 |
| T2 | 3 | 3, 0 | T3, T4 |
| T3 | 2 | 2, 0 | none |
| T4 | 3 | 1, 4 | T3 |

Schedule each step once, honor all finish-before-start links, and never exceed the listed capacities  so the service ends as early as possible.

Also, when you hand me the schedule, please use this JSON layout so I can parse it cleanly — something simple like this:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that block as a little form: each item in the ""solution"" list is one recipe step. ""task"" is the step identifier, and ""start"" and ""end"" are when that step begins and finishes (in the same time units used in the instance). It's just a sketch of the shape I expect, not your actual answer.

Please make sure you use the exact identifiers from the instance input — don't rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 4, 'nr_resources': 2, 'capacities': [4, 4], 'tasks': [{'duration': 2, 'demands': [0, 2], 'successors': [2, 1, 3]}, {'duration': 3, 'demands': [3, 0], 'successors': [2, 3]}, {'duration': 2, 'demands': [2, 0], 'successors': []}, {'duration': 3, 'demands': [1, 4], 'successors': [2]}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 2]}, {'task': 1, 'start': 2, 'end': 5, 'duration': 3, 'demands': [3, 0]}, {'task': 2, 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 3, 'start': 5, 'end': 8, 'duration': 3, 'demands': [1, 4]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': ['A', 'B'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 4}, {'resource_id': 'B', 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [0, 2], 'successors': ['T3', 'T2', 'T4']}, {'task_id': 'T2', 'duration': 3, 'demands': [3, 0], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [2, 0], 'successors': []}, {'task_id': 'T4', 'duration': 3, 'demands': [1, 4], 'successors': ['T3']}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 2]}, {'task': 'T2', 'start': 2, 'end': 5, 'duration': 3, 'demands': [3, 0]}, {'task': 'T3', 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 'T4', 'start': 5, 'end': 8, 'duration': 3, 'demands': [1, 4]}]",2,markdown_table,names
RCPSP,RCPSP,"Backstage there’s a simple rule: every scene has a fixed runtime, occupies particular camera crews and props while it runs, and some scenes depend on others finishing first. The director’s decision is to place a start time on each scene so each gets filmed exactly once, no resource is used by more scenes at the same time than exist, and all the “wait for” relationships are respected. A schedule’s quality is measured by when the last scene finishes — compute each scene’s end (start time plus duration), take the latest of those, and the goal is to make that latest finish as early as possible. The specific scene durations, resource counts and ordering details are shown below.

There are 5 scenes, 1 resource types, and resource capacities .

| resource_id | capacity |
|---|---|
| 1 | 6 |

| scene_id | scene_duration | resource_requirements | successor_scenes |
|---|---|---|---|
| T1 | 1 | 0 | T2, T4, T3, T5 |
| T2 | 3 | 3 | T3, T4, T5 |
| T3 | 3 | 5 | T5, T4 |
| T4 | 5 | 1 | T5 |
| T5 | 1 | 0 | none |

Place start times so each scene is filmed once, resources are never overused, precedences are respected, and the latest scene finish is made as early as possible.

If you want to hand me a schedule, just use this relaxed JSON layout so everything is predictable:

{
  ""solution"": [
    {
      ""task"": <scene_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little form: ""solution"" is the list of filmed scenes, each object records which scene (""task""), when it starts, and when it ends. Start and end are the numeric times (same units as the scene durations). This is only a sketch of the shape I expect — not the actual schedule.

Please use the exact scene identifiers from the instance input — no renaming, no made-up labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 5, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 3, 2, 4]}, {'duration': 3, 'demands': [3], 'successors': [2, 3, 4]}, {'duration': 3, 'demands': [5], 'successors': [4, 3]}, {'duration': 5, 'demands': [1], 'successors': [4]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 4, 'duration': 3, 'demands': [3]}, {'task': 2, 'start': 4, 'end': 7, 'duration': 3, 'demands': [5]}, {'task': 3, 'start': 7, 'end': 12, 'duration': 5, 'demands': [1]}, {'task': 4, 'start': 12, 'end': 13, 'duration': 1, 'demands': [0]}]",13.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T4', 'T3', 'T5']}, {'task_id': 'T2', 'duration': 3, 'demands': [3], 'successors': ['T3', 'T4', 'T5']}, {'task_id': 'T3', 'duration': 3, 'demands': [5], 'successors': ['T5', 'T4']}, {'task_id': 'T4', 'duration': 5, 'demands': [1], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 4, 'duration': 3, 'demands': [3]}, {'task': 'T3', 'start': 4, 'end': 7, 'duration': 3, 'demands': [5]}, {'task': 'T4', 'start': 7, 'end': 12, 'duration': 5, 'demands': [1]}, {'task': 'T5', 'start': 12, 'end': 13, 'duration': 1, 'demands': [0]}]",3,markdown_table,1
RCPSP,RCPSP,"Picture this: a release calendar with jobs that have fixed lengths, a handful of developers and a limited number of test machines, and some jobs that can’t start until others are done. The problem to solve is picking a start time for each job so that when jobs overlap they don’t use more developer hours or server slots than we’ve got, and each job is scheduled once and in the right order. The release date is simply when the last job completes, so the aim is to make that day as early as possible. The exact tasks, how long they take, their dependencies, and available resources are listed below.

# total_tasks=5
# resource_types_count=3
# resources=0, 1, 2
resource_id,capacity
0,6
1,7
2,6
task_code,task_duration,resource_demands,successor_tasks
T1,1,""1, 3, 2"",""T2, T3, T5, T4""
T2,2,""6, 2, 6"",""T5, T3, T4""
T3,5,""2, 1, 2"",""T4""
T4,1,""1, 1, 0"",""T5""
T5,1,""0, 0, 0"",""none""

Also, when you send the schedule back, just stick to a simple JSON layout so it's easy to parse. Something like this is what I expect to see:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

That's just the shape I want—""solution"" is an array of task entries. Each entry shows which task it is and the start and end times you picked for it. Keep it casual: task is the task identifier from the instance, start and end are the times (in the same time units used in the instance), and end should be after start. This block is only a sketch of the expected shape, not the actual answer.

Please make sure you use the exact identifiers from the instance input — do not rename tasks or invent new labels. Valid identifiers look like plain numbers such as ""1"" or ""23"", single capital letters like ""A"" or ""B"", or a capital letter followed by digits like ""A1"" or ""X7"".","{'nr_tasks': 5, 'nr_resources': 3, 'capacities': [6, 7, 6], 'tasks': [{'duration': 1, 'demands': [1, 3, 2], 'successors': [1, 2, 4, 3]}, {'duration': 2, 'demands': [6, 2, 6], 'successors': [4, 2, 3]}, {'duration': 5, 'demands': [2, 1, 2], 'successors': [3]}, {'duration': 1, 'demands': [1, 1, 0], 'successors': [4]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 3, 2]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [6, 2, 6]}, {'task': 2, 'start': 3, 'end': 8, 'duration': 5, 'demands': [2, 1, 2]}, {'task': 3, 'start': 8, 'end': 9, 'duration': 1, 'demands': [1, 1, 0]}, {'task': 4, 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 6}, {'resource_id': 1, 'capacity': 7}, {'resource_id': 2, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [1, 3, 2], 'successors': ['T2', 'T3', 'T5', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [6, 2, 6], 'successors': ['T5', 'T3', 'T4']}, {'task_id': 'T3', 'duration': 5, 'demands': [2, 1, 2], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 1, 'demands': [1, 1, 0], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 3, 2]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [6, 2, 6]}, {'task': 'T3', 'start': 3, 'end': 8, 'duration': 5, 'demands': [2, 1, 2]}, {'task': 'T4', 'start': 8, 'end': 9, 'duration': 1, 'demands': [1, 1, 0]}, {'task': 'T5', 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",4,csv,0
RCPSP,RCPSP,"Someone is running the event and needs to schedule every setup and program piece, knowing each one lasts a set time, uses specific volunteers and tools while active, and may be blocked until earlier pieces finish. The job is to assign start times for all items, making sure every item happens exactly once and the team never needs more volunteers or equipment than available at any instant. The winning schedule is the one that brings the whole event to a finish sooner — check the finish time of the last item, and the lower that clock reading, the better. Full details about the items and resources are shown below.

# total_items=9
# total_resource_types=2
# resources=1, 2
resource_id,capacity
1,6
2,4
item_id,duration_minutes,resource_requirements,successor_item_ids
T1,1,""0, 0"",""T2, T3, T9""
T2,2,""0, 4"",""T3, T5, T6""
T3,8,""1, 1"",""T4, T6, T7""
T4,2,""3, 0"",""T6, T7""
T5,6,""2, 0"",""T7, T6""
T6,8,""6, 0"",""T8""
T7,8,""1, 1"",""T8""
T8,1,""0, 0"",""none""
T9,3,""0, 0"",""T3""

When you send back the schedule, just use this simple JSON layout so it's easy to parse and check:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a short form: ""solution"" holds a list of all the items, and each item says which task it is and when it starts and finishes. Keep it informal — task is the item label from the event list, and start/end are the times on the event clock.

This JSON is just a sketch of the shape I need, not the actual schedule — you'll replace the placeholders with the real identifiers and numeric times.

Please make sure to use the exact identifiers from the instance input — do not rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 2, 'capacities': [6, 4], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1, 2, 8]}, {'duration': 2, 'demands': [0, 4], 'successors': [2, 4, 5]}, {'duration': 8, 'demands': [1, 1], 'successors': [3, 5, 6]}, {'duration': 2, 'demands': [3, 0], 'successors': [5, 6]}, {'duration': 6, 'demands': [2, 0], 'successors': [6, 5]}, {'duration': 8, 'demands': [6, 0], 'successors': [7]}, {'duration': 8, 'demands': [1, 1], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}, {'duration': 3, 'demands': [0, 0], 'successors': [2]}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 4]}, {'task': 2, 'start': 4, 'end': 12, 'duration': 8, 'demands': [1, 1]}, {'task': 3, 'start': 12, 'end': 14, 'duration': 2, 'demands': [3, 0]}, {'task': 4, 'start': 3, 'end': 9, 'duration': 6, 'demands': [2, 0]}, {'task': 5, 'start': 14, 'end': 22, 'duration': 8, 'demands': [6, 0]}, {'task': 6, 'start': 22, 'end': 30, 'duration': 8, 'demands': [1, 1]}, {'task': 7, 'start': 30, 'end': 31, 'duration': 1, 'demands': [0, 0]}, {'task': 8, 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 0]}]",31.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2', 'T3', 'T9']}, {'task_id': 'T2', 'duration': 2, 'demands': [0, 4], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 8, 'demands': [1, 1], 'successors': ['T4', 'T6', 'T7']}, {'task_id': 'T4', 'duration': 2, 'demands': [3, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 6, 'demands': [2, 0], 'successors': ['T7', 'T6']}, {'task_id': 'T6', 'duration': 8, 'demands': [6, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 8, 'demands': [1, 1], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}, {'task_id': 'T9', 'duration': 3, 'demands': [0, 0], 'successors': ['T3']}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 4]}, {'task': 'T3', 'start': 4, 'end': 12, 'duration': 8, 'demands': [1, 1]}, {'task': 'T4', 'start': 12, 'end': 14, 'duration': 2, 'demands': [3, 0]}, {'task': 'T5', 'start': 3, 'end': 9, 'duration': 6, 'demands': [2, 0]}, {'task': 'T6', 'start': 14, 'end': 22, 'duration': 8, 'demands': [6, 0]}, {'task': 'T7', 'start': 22, 'end': 30, 'duration': 8, 'demands': [1, 1]}, {'task': 'T8', 'start': 30, 'end': 31, 'duration': 1, 'demands': [0, 0]}, {'task': 'T9', 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 0]}]",5,csv,1
RCPSP,RCPSP,"There’s a busy workshop where every task has a known duration, uses some combination of machines and operators while it runs, and often can’t begin until a previous task is complete. The choice to make is the start time for each task so simultaneous tasks don’t exceed the available equipment or staff and so the required order of steps is preserved. A plan is judged by how quickly the last task finishes — compute each task’s finish as its start plus its run time, take the latest one, and the lower that final number the better. All tasks must be scheduled once and only once, and the specific job list and resource numbers follow below.

{
  ""total_operations"": 4,
  ""resource_types"": 2,
  ""resources"": [
    0,
    1
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 4
    },
    {
      ""resource_id"": 1,
      ""capacity"": 4
    }
  ],
  ""tasks"": [
    {
      ""operation_id"": ""T1"",
      ""processing_time"": 2,
      ""resource_requirements"": [
        0,
        4
      ],
      ""immediate_successors"": [
        ""T3"",
        ""T2"",
        ""T4""
      ]
    },
    {
      ""operation_id"": ""T2"",
      ""processing_time"": 1,
      ""resource_requirements"": [
        4,
        0
      ],
      ""immediate_successors"": [
        ""T3"",
        ""T4""
      ]
    },
    {
      ""operation_id"": ""T3"",
      ""processing_time"": 4,
      ""resource_requirements"": [
        3,
        0
      ],
      ""immediate_successors"": [
        ""T4""
      ]
    },
    {
      ""operation_id"": ""T4"",
      ""processing_time"": 5,
      ""resource_requirements"": [
        1,
        2
      ],
      ""immediate_successors"": []
    }
  ]
}

You can return the schedule in a simple JSON shape — nothing fancy — just a list of tasks with their start and end times, for example:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This shows the structure I expect: ""solution"" is an array of entries, and each entry is a task record with the task identifier plus the start and end times (end would be the start plus the task's duration). Treat it like a little form for each job.

Note that the JSON above is just a sketch of the expected shape — not the actual schedule. When you submit your answer, fill in the real task identifiers and times.

All identifiers must be used exactly as they appear in the instance input — no renaming and no new labels.
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 4, 'nr_resources': 2, 'capacities': [4, 4], 'tasks': [{'duration': 2, 'demands': [0, 4], 'successors': [2, 1, 3]}, {'duration': 1, 'demands': [4, 0], 'successors': [2, 3]}, {'duration': 4, 'demands': [3, 0], 'successors': [3]}, {'duration': 5, 'demands': [1, 2], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 4]}, {'task': 1, 'start': 2, 'end': 3, 'duration': 1, 'demands': [4, 0]}, {'task': 2, 'start': 3, 'end': 7, 'duration': 4, 'demands': [3, 0]}, {'task': 3, 'start': 7, 'end': 12, 'duration': 5, 'demands': [1, 2]}]",12.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 4}, {'resource_id': 1, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [0, 4], 'successors': ['T3', 'T2', 'T4']}, {'task_id': 'T2', 'duration': 1, 'demands': [4, 0], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 4, 'demands': [3, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 5, 'demands': [1, 2], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 4]}, {'task': 'T2', 'start': 2, 'end': 3, 'duration': 1, 'demands': [4, 0]}, {'task': 'T3', 'start': 3, 'end': 7, 'duration': 4, 'demands': [3, 0]}, {'task': 'T4', 'start': 7, 'end': 12, 'duration': 5, 'demands': [1, 2]}]",6,json,0
RCPSP,RCPSP,"Picture this: the head nurse has a list of fixed-time procedures, each of which consumes one or more operating rooms and staff members while it runs, and several procedures are chained so one can’t start until another has finished. The task is simply choosing start times so no rooms or staff are double-booked, every dependency is honored, and every procedure is scheduled once. A clearer plan is the one where the whole set of procedures is done sooner — find each procedure’s end by adding its duration to its start, then take the latest of those end times; a smaller latest end time means a better schedule. The precise patient procedures, their lengths, resource capacities, and ordering constraints are shown below.

{
  ""num_procedures"": 4,
  ""num_resource_types"": 3,
  ""resources"": [
    ""A"",
    ""B"",
    ""C""
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": ""A"",
      ""capacity"": 4
    },
    {
      ""resource_id"": ""B"",
      ""capacity"": 7
    },
    {
      ""resource_id"": ""C"",
      ""capacity"": 5
    }
  ],
  ""tasks"": [
    {
      ""procedure_id"": ""T1"",
      ""procedure_duration"": 1,
      ""resource_demands"": [
        3,
        2,
        1
      ],
      ""successor_procedures"": [
        ""T2"",
        ""T3"",
        ""T4""
      ]
    },
    {
      ""procedure_id"": ""T2"",
      ""procedure_duration"": 2,
      ""resource_demands"": [
        2,
        1,
        5
      ],
      ""successor_procedures"": [
        ""T3"",
        ""T4""
      ]
    },
    {
      ""procedure_id"": ""T3"",
      ""procedure_duration"": 4,
      ""resource_demands"": [
        1,
        1,
        0
      ],
      ""successor_procedures"": [
        ""T4""
      ]
    },
    {
      ""procedure_id"": ""T4"",
      ""procedure_duration"": 1,
      ""resource_demands"": [
        0,
        0,
        0
      ],
      ""successor_procedures"": []
    }
  ]
}

Also, to keep things machine-friendly and easy to check, please return the schedule using this simple JSON layout:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

It's just a little form: ""solution"" is a list, and each entry has the task identifier plus its start and end times. Think of each entry like a line on the schedule — which procedure, when it starts, and when it finishes. This is just the shape I expect, not the actual filled-in schedule.

Please make sure to use the exact task identifiers from the instance input — do not rename them or introduce new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 4, 'nr_resources': 3, 'capacities': [4, 7, 5], 'tasks': [{'duration': 1, 'demands': [3, 2, 1], 'successors': [1, 2, 3]}, {'duration': 2, 'demands': [2, 1, 5], 'successors': [2, 3]}, {'duration': 4, 'demands': [1, 1, 0], 'successors': [3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [3, 2, 1]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [2, 1, 5]}, {'task': 2, 'start': 3, 'end': 7, 'duration': 4, 'demands': [1, 1, 0]}, {'task': 3, 'start': 7, 'end': 8, 'duration': 1, 'demands': [0, 0, 0]}]",8.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 4}, {'resource_id': 'B', 'capacity': 7}, {'resource_id': 'C', 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [3, 2, 1], 'successors': ['T2', 'T3', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [2, 1, 5], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 4, 'demands': [1, 1, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [3, 2, 1]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [2, 1, 5]}, {'task': 'T3', 'start': 3, 'end': 7, 'duration': 4, 'demands': [1, 1, 0]}, {'task': 'T4', 'start': 7, 'end': 8, 'duration': 1, 'demands': [0, 0, 0]}]",7,json,names
RCPSP,RCPSP,"Recently the foreman sat down with a whiteboard of repairs: fixed-duration jobs, each consuming some cranes and crews while active, and a few that can’t start until others are completed. The work is to place every job on the timeline exactly once, respect those “finish-before” links, and never let the running jobs demand more cranes or people than are on hand. The straightforward measure of a good plan is when the last job is completed — the earlier that moment, the better. The exact jobs, durations and available cranes/crews are listed below.

# total_jobs_count=4
# resource_types_count=2
# resources=1, 2
resource_id,capacity
1,5
2,5
job_id,job_duration,resource_demands,successor_job_ids
T1,3,""0, 5"",""T3, T4""
T2,1,""3, 0"",""T3, T4, T1""
T3,2,""5, 0"",""T4""
T4,6,""1, 2"",""none""

Oh, and when you're ready to hand the schedule back, just use this little JSON shape so it's easy to parse and check:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of each object in the ""solution"" list as one job on the timeline: the task identifier, when it starts, and when it finishes. It's just a simple form — list every task once with its start and end times. This block is only a sketch of the expected shape, not the actual filled-in schedule.

Please use the exact task identifiers from the instance input — don't rename them and don't invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 4, 'nr_resources': 2, 'capacities': [5, 5], 'tasks': [{'duration': 3, 'demands': [0, 5], 'successors': [2, 3]}, {'duration': 1, 'demands': [3, 0], 'successors': [2, 3, 0]}, {'duration': 2, 'demands': [5, 0], 'successors': [3]}, {'duration': 6, 'demands': [1, 2], 'successors': []}]}","[{'task': 0, 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 5]}, {'task': 1, 'start': 0, 'end': 1, 'duration': 1, 'demands': [3, 0]}, {'task': 2, 'start': 4, 'end': 6, 'duration': 2, 'demands': [5, 0]}, {'task': 3, 'start': 6, 'end': 12, 'duration': 6, 'demands': [1, 2]}]",12.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 3, 'demands': [0, 5], 'successors': ['T3', 'T4']}, {'task_id': 'T2', 'duration': 1, 'demands': [3, 0], 'successors': ['T3', 'T4', 'T1']}, {'task_id': 'T3', 'duration': 2, 'demands': [5, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 6, 'demands': [1, 2], 'successors': []}]}","[{'task': 'T1', 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 5]}, {'task': 'T2', 'start': 0, 'end': 1, 'duration': 1, 'demands': [3, 0]}, {'task': 'T3', 'start': 4, 'end': 6, 'duration': 2, 'demands': [5, 0]}, {'task': 'T4', 'start': 6, 'end': 12, 'duration': 6, 'demands': [1, 2]}]",8,csv,1
RCPSP,RCPSP,"Recently the team took on a landscaping overhaul where every step has a set duration, ties up a certain number of tractors and crew while underway, and depends on earlier prep in many cases. The coordinator’s role is to set start times for every step so all dependencies are honored, the number of tractors and crew being used at once never exceeds what’s available, and each step happens once. To compare different plans, just check when the final step ends — the sooner the entire job is done, the better. The detailed list of steps, times, and available tractors/crew appears below.

There are 5 landscaping steps, 1 resource types, and the available resources are .
Step T1 runs for 1 time units, uses resources 0, and has successors T2, T4, T3, T5 that must start after it finishes.
Step T2 runs for 5 time units, uses resources 3, and has successors T3, T5, T4 that must start after it finishes.
Step T3 runs for 3 time units, uses resources 6, and has successors T5, T4 that must start after it finishes.
Step T4 runs for 2 time units, uses resources 2, and has successors T5 that must start after it finishes.
Step T5 runs for 1 time units, uses resources 0, and has successors none that must start after it finishes.
The coordinator must assign start times so all 5 steps are scheduled, the 1 resource capacities  are never exceeded, and the overall completion time is minimized.

When you hand the schedule back, a simple JSON layout is perfect — nothing fancy, just the fields I showed below so it's easy to parse.

{
  ""solution"": [
    {
      ""task"": <step_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of this like a short form: ""task"" is the step identifier from the job list, ""start"" is when that step kicks off, and ""end"" is when it finishes. Super casual — just fill in each step's ID and times. This block is only a sketch of the shape I expect, not the actual filled-in answer.

Please use the identifiers exactly as they appear in the instance input — no renaming and no new labels.

Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 5, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 3, 2, 4]}, {'duration': 5, 'demands': [3], 'successors': [2, 4, 3]}, {'duration': 3, 'demands': [6], 'successors': [4, 3]}, {'duration': 2, 'demands': [2], 'successors': [4]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 6, 'duration': 5, 'demands': [3]}, {'task': 2, 'start': 6, 'end': 9, 'duration': 3, 'demands': [6]}, {'task': 3, 'start': 9, 'end': 11, 'duration': 2, 'demands': [2]}, {'task': 4, 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",12.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T4', 'T3', 'T5']}, {'task_id': 'T2', 'duration': 5, 'demands': [3], 'successors': ['T3', 'T5', 'T4']}, {'task_id': 'T3', 'duration': 3, 'demands': [6], 'successors': ['T5', 'T4']}, {'task_id': 'T4', 'duration': 2, 'demands': [2], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 6, 'duration': 5, 'demands': [3]}, {'task': 'T3', 'start': 6, 'end': 9, 'duration': 3, 'demands': [6]}, {'task': 'T4', 'start': 9, 'end': 11, 'duration': 2, 'demands': [2]}, {'task': 'T5', 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",9,nl,1
RCPSP,RCPSP,"Someone managing the shop needs to line up every assembly step like dominos. Each step lasts a set period, uses a given number of tools and a patch of bench while active, and some steps can only start after others finish. The plan must assign a start time to every step (no skipping, no double-running), keep simultaneous tool and bench usage within what’s available, and aim to get the final piece finished as soon as possible — measured by the finishing time of the last step, with smaller times better. The specific tasks, durations, and available tools/benches appear below.

{
  ""total_assembly_steps"": 8,
  ""resource_types_count"": 2,
  ""resources"": [
    0,
    1
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 3
    },
    {
      ""resource_id"": 1,
      ""capacity"": 4
    }
  ],
  ""tasks"": [
    {
      ""step_id"": ""T1"",
      ""step_duration"": 1,
      ""step_resource_usage"": [
        0,
        0
      ],
      ""following_steps"": [
        ""T2""
      ]
    },
    {
      ""step_id"": ""T2"",
      ""step_duration"": 4,
      ""step_resource_usage"": [
        0,
        3
      ],
      ""following_steps"": [
        ""T3"",
        ""T6""
      ]
    },
    {
      ""step_id"": ""T3"",
      ""step_duration"": 7,
      ""step_resource_usage"": [
        1,
        2
      ],
      ""following_steps"": []
    },
    {
      ""step_id"": ""T4"",
      ""step_duration"": 3,
      ""step_resource_usage"": [
        3,
        0
      ],
      ""following_steps"": [
        ""T6"",
        ""T7""
      ]
    },
    {
      ""step_id"": ""T5"",
      ""step_duration"": 12,
      ""step_resource_usage"": [
        1,
        0
      ],
      ""following_steps"": [
        ""T7""
      ]
    },
    {
      ""step_id"": ""T6"",
      ""step_duration"": 6,
      ""step_resource_usage"": [
        2,
        0
      ],
      ""following_steps"": [
        ""T8""
      ]
    },
    {
      ""step_id"": ""T7"",
      ""step_duration"": 5,
      ""step_resource_usage"": [
        1,
        3
      ],
      ""following_steps"": [
        ""T8""
      ]
    },
    {
      ""step_id"": ""T8"",
      ""step_duration"": 1,
      ""step_resource_usage"": [
        0,
        0
      ],
      ""following_steps"": []
    }
  ]
}

Also, when you send the scheduled plan back, a simple JSON sketch like this is the format I expect — just showing the shape of the answer so it's easy to parse:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little form: each object in ""solution"" is one assembly step, ""task"" picks which step, and ""start"" and ""end"" say when it begins and finishes. This is just the shape I want you to follow — not the actual schedule itself.

Also, please use the exact task identifiers from the instance input — don't rename tasks or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 8, 'nr_resources': 2, 'capacities': [3, 4], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1]}, {'duration': 4, 'demands': [0, 3], 'successors': [2, 5]}, {'duration': 7, 'demands': [1, 2], 'successors': []}, {'duration': 3, 'demands': [3, 0], 'successors': [5, 6]}, {'duration': 12, 'demands': [1, 0], 'successors': [6]}, {'duration': 6, 'demands': [2, 0], 'successors': [7]}, {'duration': 5, 'demands': [1, 3], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 3]}, {'task': 2, 'start': 5, 'end': 12, 'duration': 7, 'demands': [1, 2]}, {'task': 3, 'start': 0, 'end': 3, 'duration': 3, 'demands': [3, 0]}, {'task': 4, 'start': 3, 'end': 15, 'duration': 12, 'demands': [1, 0]}, {'task': 5, 'start': 12, 'end': 18, 'duration': 6, 'demands': [2, 0]}, {'task': 6, 'start': 15, 'end': 20, 'duration': 5, 'demands': [1, 3]}, {'task': 7, 'start': 20, 'end': 21, 'duration': 1, 'demands': [0, 0]}]",21.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 3}, {'resource_id': 1, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 4, 'demands': [0, 3], 'successors': ['T3', 'T6']}, {'task_id': 'T3', 'duration': 7, 'demands': [1, 2], 'successors': []}, {'task_id': 'T4', 'duration': 3, 'demands': [3, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 12, 'demands': [1, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 6, 'demands': [2, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 5, 'demands': [1, 3], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 3]}, {'task': 'T3', 'start': 5, 'end': 12, 'duration': 7, 'demands': [1, 2]}, {'task': 'T4', 'start': 0, 'end': 3, 'duration': 3, 'demands': [3, 0]}, {'task': 'T5', 'start': 3, 'end': 15, 'duration': 12, 'demands': [1, 0]}, {'task': 'T6', 'start': 12, 'end': 18, 'duration': 6, 'demands': [2, 0]}, {'task': 'T7', 'start': 15, 'end': 20, 'duration': 5, 'demands': [1, 3]}, {'task': 'T8', 'start': 20, 'end': 21, 'duration': 1, 'demands': [0, 0]}]",10,json,0
RCPSP,RCPSP,"Back when the new term calendar landed on my desk, it became clear: every classroom setup and lesson has a fixed length, uses up rooms and teachers while active, and some can’t begin until others have wrapped. The job is to pick a start time for every single item so we never have more concurrent classes than rooms or teachers available, and nothing is skipped or scheduled twice. The goal is to get everything done as soon as possible — in other words, make the time the final session ends as small as it can be. The concrete session and resource details will be shown below.

# nr_sessions=8
# nr_resource_types=2
# resources=0, 1
resource_id,capacity
0,4
1,6
session_id,session_duration,resource_requirements,successor_sessions
T1,1,""0, 0"",""T2""
T2,4,""0, 6"",""T3, T5, T6""
T3,3,""2, 2"",""T4, T6""
T4,2,""2, 0"",""T6, T7""
T5,4,""1, 0"",""T7""
T6,2,""2, 0"",""T8""
T7,4,""2, 2"",""T8""
T8,1,""0, 0"",""none""

Oh, and when you send the scheduled plan back, keep it in a tiny JSON layout so it's easy to check automatically — nothing fancy, just this shape:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This shows the shape I expect: ""solution"" is a list of scheduled items, and each item has a task identifier plus the start and end times for that session. Treat it like a simple form — each row says which session it is and when it begins and finishes. The block above is just a sketch of the expected shape, not the actual answer — replace those placeholders with the real IDs and times from the instance.

Please use the exact identifiers from the input — do not rename them or invent new labels. 
Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 8, 'nr_resources': 2, 'capacities': [4, 6], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1]}, {'duration': 4, 'demands': [0, 6], 'successors': [2, 4, 5]}, {'duration': 3, 'demands': [2, 2], 'successors': [3, 5]}, {'duration': 2, 'demands': [2, 0], 'successors': [5, 6]}, {'duration': 4, 'demands': [1, 0], 'successors': [6]}, {'duration': 2, 'demands': [2, 0], 'successors': [7]}, {'duration': 4, 'demands': [2, 2], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 6]}, {'task': 2, 'start': 5, 'end': 8, 'duration': 3, 'demands': [2, 2]}, {'task': 3, 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 4, 'start': 5, 'end': 9, 'duration': 4, 'demands': [1, 0]}, {'task': 5, 'start': 10, 'end': 12, 'duration': 2, 'demands': [2, 0]}, {'task': 6, 'start': 10, 'end': 14, 'duration': 4, 'demands': [2, 2]}, {'task': 7, 'start': 14, 'end': 15, 'duration': 1, 'demands': [0, 0]}]",15.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 4}, {'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 4, 'demands': [0, 6], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 3, 'demands': [2, 2], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 2, 'demands': [2, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 4, 'demands': [1, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 2, 'demands': [2, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 4, 'demands': [2, 2], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 6]}, {'task': 'T3', 'start': 5, 'end': 8, 'duration': 3, 'demands': [2, 2]}, {'task': 'T4', 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 'T5', 'start': 5, 'end': 9, 'duration': 4, 'demands': [1, 0]}, {'task': 'T6', 'start': 10, 'end': 12, 'duration': 2, 'demands': [2, 0]}, {'task': 'T7', 'start': 10, 'end': 14, 'duration': 4, 'demands': [2, 2]}, {'task': 'T8', 'start': 14, 'end': 15, 'duration': 1, 'demands': [0, 0]}]",11,csv,0
RCPSP,RCPSP,"Recently the photographer picked up a complicated job: several fixed-length shoot segments, limited lighting rigs and crew, and a bunch of “must-finish-before” relationships between segments. The job now is to decide when each segment starts so no two simultaneous segments together demand more gear or staff than exist, and so each segment is done exactly once in its required order. Better schedules are the ones that get the last segment done earlier — the finish of the shoot is simply the end time of that last segment, and the aim is to push that end time as early as possible. The detailed list of segments, their durations and the resource availability is listed below.

The detailed list follows: 9 total segments, 3 resource types, capacities .
Segment T1: duration 1, demands 0, 0, 0 while running, must finish before T2, T3, T7.
Segment T2: duration 2, demands 3, 4, 1 while running, must finish before T4, T5.
Segment T3: duration 2, demands 0, 3, 1 while running, must finish before T6, T4.
Segment T4: duration 3, demands 2, 3, 3 while running, must finish before T7.
Segment T5: duration 3, demands 1, 1, 4 while running, must finish before T6, T4.
Segment T6: duration 2, demands 1, 2, 0 while running, must finish before T7, T4.
Segment T7: duration 1, demands 0, 0, 0 while running, must finish before none.
Segment T8: duration 3, demands 2, 3, 4 while running, must finish before T7, T4.
Segment T9: duration 2, demands 0, 0, 0 while running, must finish before T6, T8.
All segments must be scheduled once respecting these precedences and resource capacities so the shoot finishes as early as possible.

When you’re ready to give me the schedule, a simple JSON layout like this works best — it's just a small list of entries, each saying which segment (task) starts when and ends when.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little form: ""solution"" holds a list of scheduled segments, each item gives the segment's id under ""task"" and the time it starts and the time it finishes under ""start"" and ""end"". Keep it light — just one object per segment, and the end time should match when that segment is done.

This JSON is only a sketch of the shape I expect, not the actual filled-in schedule.

Also, please use the exact identifiers from the instance input — don’t rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 3, 'capacities': [6, 5, 4], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2, 6]}, {'duration': 2, 'demands': [3, 4, 1], 'successors': [3, 4]}, {'duration': 2, 'demands': [0, 3, 1], 'successors': [5, 3]}, {'duration': 3, 'demands': [2, 3, 3], 'successors': [6]}, {'duration': 3, 'demands': [1, 1, 4], 'successors': [5, 3]}, {'duration': 2, 'demands': [1, 2, 0], 'successors': [6, 3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 3, 'demands': [2, 3, 4], 'successors': [6, 3]}, {'duration': 2, 'demands': [0, 0, 0], 'successors': [5, 7]}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [3, 4, 1]}, {'task': 2, 'start': 6, 'end': 8, 'duration': 2, 'demands': [0, 3, 1]}, {'task': 3, 'start': 11, 'end': 14, 'duration': 3, 'demands': [2, 3, 3]}, {'task': 4, 'start': 3, 'end': 6, 'duration': 3, 'demands': [1, 1, 4]}, {'task': 5, 'start': 8, 'end': 10, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 6, 'start': 14, 'end': 15, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 3, 4]}, {'task': 8, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 0, 0]}]",15.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': [1, 2, 3], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 5}, {'resource_id': 3, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3', 'T7']}, {'task_id': 'T2', 'duration': 2, 'demands': [3, 4, 1], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 2, 'demands': [0, 3, 1], 'successors': ['T6', 'T4']}, {'task_id': 'T4', 'duration': 3, 'demands': [2, 3, 3], 'successors': ['T7']}, {'task_id': 'T5', 'duration': 3, 'demands': [1, 1, 4], 'successors': ['T6', 'T4']}, {'task_id': 'T6', 'duration': 2, 'demands': [1, 2, 0], 'successors': ['T7', 'T4']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 3, 'demands': [2, 3, 4], 'successors': ['T7', 'T4']}, {'task_id': 'T9', 'duration': 2, 'demands': [0, 0, 0], 'successors': ['T6', 'T8']}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [3, 4, 1]}, {'task': 'T3', 'start': 6, 'end': 8, 'duration': 2, 'demands': [0, 3, 1]}, {'task': 'T4', 'start': 11, 'end': 14, 'duration': 3, 'demands': [2, 3, 3]}, {'task': 'T5', 'start': 3, 'end': 6, 'duration': 3, 'demands': [1, 1, 4]}, {'task': 'T6', 'start': 8, 'end': 10, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 'T7', 'start': 14, 'end': 15, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 3, 4]}, {'task': 'T9', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 0, 0]}]",12,nl,1
RCPSP,RCPSP,"Someone on the production team mapped out all the songs: each has a known recording length, needs specific studio resources while it’s happening, and some songs depend on others being completed first. The task becomes assigning a start time to each of those sessions so there are never more overlapping sessions than studios or engineers allow, and every required session happens once and only once. What makes one schedule better than another is simple — it finishes the album earlier; the album’s completion time is the time the final session wraps up, and that finish time is what to watch. The concrete numbers and constraints are shown below.

{
  ""total_sessions"": 7,
  ""resource_types_count"": 2,
  ""resources"": [
    0,
    1
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 3
    },
    {
      ""resource_id"": 1,
      ""capacity"": 4
    }
  ],
  ""tasks"": [
    {
      ""session_id"": ""T1"",
      ""session_duration"": 2,
      ""session_resource_needs"": [
        0,
        3
      ],
      ""must_follow_sessions"": [
        ""T2"",
        ""T4"",
        ""T5""
      ]
    },
    {
      ""session_id"": ""T2"",
      ""session_duration"": 6,
      ""session_resource_needs"": [
        2,
        4
      ],
      ""must_follow_sessions"": [
        ""T3"",
        ""T5"",
        ""T6""
      ]
    },
    {
      ""session_id"": ""T3"",
      ""session_duration"": 2,
      ""session_resource_needs"": [
        2,
        0
      ],
      ""must_follow_sessions"": [
        ""T5"",
        ""T6""
      ]
    },
    {
      ""session_id"": ""T4"",
      ""session_duration"": 14,
      ""session_resource_needs"": [
        1,
        0
      ],
      ""must_follow_sessions"": [
        ""T6""
      ]
    },
    {
      ""session_id"": ""T5"",
      ""session_duration"": 4,
      ""session_resource_needs"": [
        3,
        0
      ],
      ""must_follow_sessions"": [
        ""T7""
      ]
    },
    {
      ""session_id"": ""T6"",
      ""session_duration"": 5,
      ""session_resource_needs"": [
        1,
        3
      ],
      ""must_follow_sessions"": [
        ""T7""
      ]
    },
    {
      ""session_id"": ""T7"",
      ""session_duration"": 1,
      ""session_resource_needs"": [
        0,
        0
      ],
      ""must_follow_sessions"": []
    }
  ]
}

Also, when you send the actual schedule back, please use this JSON layout so it's easy to read and check:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of this like a simple form: ""solution"" is the list of all recording sessions, and each item says which session it is (""task"") and when it starts and finishes (""start"" and ""end""). Keep it casual — task is the song/session identifier, start and end are the times (same units as the instance) when that session begins and ends.

This JSON is just a sketch of the shape I expect, not the final scheduled times.

Please make sure you use the exact identifiers from the instance input — do not rename them or invent new ones.  
For example: Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 7, 'nr_resources': 2, 'capacities': [3, 4], 'tasks': [{'duration': 2, 'demands': [0, 3], 'successors': [1, 3, 4]}, {'duration': 6, 'demands': [2, 4], 'successors': [2, 4, 5]}, {'duration': 2, 'demands': [2, 0], 'successors': [4, 5]}, {'duration': 14, 'demands': [1, 0], 'successors': [5]}, {'duration': 4, 'demands': [3, 0], 'successors': [6]}, {'duration': 5, 'demands': [1, 3], 'successors': [6]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 3]}, {'task': 1, 'start': 2, 'end': 8, 'duration': 6, 'demands': [2, 4]}, {'task': 2, 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 3, 'start': 2, 'end': 16, 'duration': 14, 'demands': [1, 0]}, {'task': 4, 'start': 21, 'end': 25, 'duration': 4, 'demands': [3, 0]}, {'task': 5, 'start': 16, 'end': 21, 'duration': 5, 'demands': [1, 3]}, {'task': 6, 'start': 25, 'end': 26, 'duration': 1, 'demands': [0, 0]}]",26.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 3}, {'resource_id': 1, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [0, 3], 'successors': ['T2', 'T4', 'T5']}, {'task_id': 'T2', 'duration': 6, 'demands': [2, 4], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 2, 'demands': [2, 0], 'successors': ['T5', 'T6']}, {'task_id': 'T4', 'duration': 14, 'demands': [1, 0], 'successors': ['T6']}, {'task_id': 'T5', 'duration': 4, 'demands': [3, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 5, 'demands': [1, 3], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 3]}, {'task': 'T2', 'start': 2, 'end': 8, 'duration': 6, 'demands': [2, 4]}, {'task': 'T3', 'start': 8, 'end': 10, 'duration': 2, 'demands': [2, 0]}, {'task': 'T4', 'start': 2, 'end': 16, 'duration': 14, 'demands': [1, 0]}, {'task': 'T5', 'start': 21, 'end': 25, 'duration': 4, 'demands': [3, 0]}, {'task': 'T6', 'start': 16, 'end': 21, 'duration': 5, 'demands': [1, 3]}, {'task': 'T7', 'start': 25, 'end': 26, 'duration': 1, 'demands': [0, 0]}]",13,json,0
RCPSP,RCPSP,"Someone in the lab needs to map out the day: every experiment has a fixed runtime, it uses up certain instruments and tech time while it’s active, and some experiments depend on the results of earlier ones. The job is to pick start times, making sure each experiment runs exactly once, dependencies are honored, and at no point do equipment or technicians get overbooked. What makes one timetable better than another is simple — the sooner the last experiment is done, the better; use that final completion time to compare plans. The specific experiment schedule and resource limits are given below.

There are 7 experiments to schedule, with 3 resource types available and capacities .

| resource_id | capacity |
|---|---|
| A | 2 |
| B | 2 |
| C | 3 |

| experiment_id | runtime | resource_demands | dependent_experiments |
|---|---|---|---|
| T1 | 8 | 0, 0, 0 | T2, T4, T3 |
| T2 | 2 | 1, 0, 0 | T3, T5 |
| T3 | 1 | 0, 0, 0 | T7, T5, T6 |
| T4 | 4 | 0, 0, 1 | T6, T5 |
| T5 | 2 | 0, 0, 1 | T6 |
| T6 | 5 | 0, 2, 0 | T7 |
| T7 | 8 | 0, 0, 0 | none |

Respect these experiment runtimes, resource demands, and precedences; use the final completion time to compare timetables.

Also, when you hand me the timetable, please use a simple JSON shape so it's easy to parse and check.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little form: ""solution"" holds a list of experiment entries; each entry names the task (the experiment label) and the start and end times when that experiment runs. It's just a sketch of the shape I expect, not the actual schedule — you'll fill in the real task ids and times.

One more important thing: use the exact identifiers from the instance input — don't rename them or invent new ones. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 7, 'nr_resources': 3, 'capacities': [2, 2, 3], 'tasks': [{'duration': 8, 'demands': [0, 0, 0], 'successors': [1, 3, 2]}, {'duration': 2, 'demands': [1, 0, 0], 'successors': [2, 4]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': [6, 4, 5]}, {'duration': 4, 'demands': [0, 0, 1], 'successors': [5, 4]}, {'duration': 2, 'demands': [0, 0, 1], 'successors': [5]}, {'duration': 5, 'demands': [0, 2, 0], 'successors': [6]}, {'duration': 8, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 8, 'duration': 8, 'demands': [0, 0, 0]}, {'task': 1, 'start': 8, 'end': 10, 'duration': 2, 'demands': [1, 0, 0]}, {'task': 2, 'start': 11, 'end': 12, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 3, 'start': 8, 'end': 12, 'duration': 4, 'demands': [0, 0, 1]}, {'task': 4, 'start': 12, 'end': 14, 'duration': 2, 'demands': [0, 0, 1]}, {'task': 5, 'start': 14, 'end': 19, 'duration': 5, 'demands': [0, 2, 0]}, {'task': 6, 'start': 19, 'end': 27, 'duration': 8, 'demands': [0, 0, 0]}]",27.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 2}, {'resource_id': 'B', 'capacity': 2}, {'resource_id': 'C', 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 8, 'demands': [0, 0, 0], 'successors': ['T2', 'T4', 'T3']}, {'task_id': 'T2', 'duration': 2, 'demands': [1, 0, 0], 'successors': ['T3', 'T5']}, {'task_id': 'T3', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T7', 'T5', 'T6']}, {'task_id': 'T4', 'duration': 4, 'demands': [0, 0, 1], 'successors': ['T6', 'T5']}, {'task_id': 'T5', 'duration': 2, 'demands': [0, 0, 1], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 5, 'demands': [0, 2, 0], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 8, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 8, 'duration': 8, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 8, 'end': 10, 'duration': 2, 'demands': [1, 0, 0]}, {'task': 'T3', 'start': 11, 'end': 12, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T4', 'start': 8, 'end': 12, 'duration': 4, 'demands': [0, 0, 1]}, {'task': 'T5', 'start': 12, 'end': 14, 'duration': 2, 'demands': [0, 0, 1]}, {'task': 'T6', 'start': 14, 'end': 19, 'duration': 5, 'demands': [0, 2, 0]}, {'task': 'T7', 'start': 19, 'end': 27, 'duration': 8, 'demands': [0, 0, 0]}]",14,markdown_table,names
RCPSP,RCPSP,"Someone on the floor is juggling start times for each picking and packing job — each job takes a set amount of time, needs a few forklifts and some hands while it’s running, and often has to wait for other jobs to finish first. The plan must include every job exactly once, keep the running count of forklifts and staff within the limits at all moments, and try to make the shipment go out as early as possible — that target is simply the time when the last job ends. The detailed instance (jobs, times, and resources) appears below.

{
  ""total_jobs"": 10,
  ""resource_types_count"": 1,
  ""resources"": [
    0
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 5
    }
  ],
  ""tasks"": [
    {
      ""job_id"": ""T1"",
      ""job_duration"": 1,
      ""resource_demands"": [
        0
      ],
      ""successor_jobs"": [
        ""T2"",
        ""T3"",
        ""T8""
      ]
    },
    {
      ""job_id"": ""T2"",
      ""job_duration"": 3,
      ""resource_demands"": [
        3
      ],
      ""successor_jobs"": [
        ""T5""
      ]
    },
    {
      ""job_id"": ""T3"",
      ""job_duration"": 6,
      ""resource_demands"": [
        5
      ],
      ""successor_jobs"": [
        ""T6""
      ]
    },
    {
      ""job_id"": ""T4"",
      ""job_duration"": 1,
      ""resource_demands"": [
        3
      ],
      ""successor_jobs"": []
    },
    {
      ""job_id"": ""T5"",
      ""job_duration"": 2,
      ""resource_demands"": [
        3
      ],
      ""successor_jobs"": [
        ""T9""
      ]
    },
    {
      ""job_id"": ""T6"",
      ""job_duration"": 3,
      ""resource_demands"": [
        2
      ],
      ""successor_jobs"": [
        ""T7""
      ]
    },
    {
      ""job_id"": ""T7"",
      ""job_duration"": 4,
      ""resource_demands"": [
        1
      ],
      ""successor_jobs"": [
        ""T9""
      ]
    },
    {
      ""job_id"": ""T8"",
      ""job_duration"": 4,
      ""resource_demands"": [
        1
      ],
      ""successor_jobs"": []
    },
    {
      ""job_id"": ""T9"",
      ""job_duration"": 1,
      ""resource_demands"": [
        0
      ],
      ""successor_jobs"": []
    },
    {
      ""job_id"": ""T10"",
      ""job_duration"": 2,
      ""resource_demands"": [
        3
      ],
      ""successor_jobs"": []
    }
  ]
}

If you want to hand me the schedule, a nice, tidy JSON layout works best — something like this shape will do the job:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just sketches the shape: ""solution"" is the list of all jobs, and each entry says which task (its id) and when it starts and ends. Think of it like filling out a little form for each picking/packing job: the job label, its start time, and its end time. It’s just an example of the format I expect, not the actual schedule.

Please make sure to use the exact identifiers from the instance input — don’t rename tasks or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 10, 'nr_resources': 1, 'capacities': [5], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 2, 7]}, {'duration': 3, 'demands': [3], 'successors': [4]}, {'duration': 6, 'demands': [5], 'successors': [5]}, {'duration': 1, 'demands': [3], 'successors': []}, {'duration': 2, 'demands': [3], 'successors': [8]}, {'duration': 3, 'demands': [2], 'successors': [6]}, {'duration': 4, 'demands': [1], 'successors': [8]}, {'duration': 4, 'demands': [1], 'successors': []}, {'duration': 1, 'demands': [0], 'successors': []}, {'duration': 2, 'demands': [3], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 7, 'end': 10, 'duration': 3, 'demands': [3]}, {'task': 2, 'start': 1, 'end': 7, 'duration': 6, 'demands': [5]}, {'task': 3, 'start': 0, 'end': 1, 'duration': 1, 'demands': [3]}, {'task': 4, 'start': 10, 'end': 12, 'duration': 2, 'demands': [3]}, {'task': 5, 'start': 7, 'end': 10, 'duration': 3, 'demands': [2]}, {'task': 6, 'start': 10, 'end': 14, 'duration': 4, 'demands': [1]}, {'task': 7, 'start': 10, 'end': 14, 'duration': 4, 'demands': [1]}, {'task': 8, 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}, {'task': 9, 'start': 12, 'end': 14, 'duration': 2, 'demands': [3]}]",15.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [0], 'resource_capacities': [{'resource_id': 0, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T3', 'T8']}, {'task_id': 'T2', 'duration': 3, 'demands': [3], 'successors': ['T5']}, {'task_id': 'T3', 'duration': 6, 'demands': [5], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 1, 'demands': [3], 'successors': []}, {'task_id': 'T5', 'duration': 2, 'demands': [3], 'successors': ['T9']}, {'task_id': 'T6', 'duration': 3, 'demands': [2], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 4, 'demands': [1], 'successors': ['T9']}, {'task_id': 'T8', 'duration': 4, 'demands': [1], 'successors': []}, {'task_id': 'T9', 'duration': 1, 'demands': [0], 'successors': []}, {'task_id': 'T10', 'duration': 2, 'demands': [3], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 7, 'end': 10, 'duration': 3, 'demands': [3]}, {'task': 'T3', 'start': 1, 'end': 7, 'duration': 6, 'demands': [5]}, {'task': 'T4', 'start': 0, 'end': 1, 'duration': 1, 'demands': [3]}, {'task': 'T5', 'start': 10, 'end': 12, 'duration': 2, 'demands': [3]}, {'task': 'T6', 'start': 7, 'end': 10, 'duration': 3, 'demands': [2]}, {'task': 'T7', 'start': 10, 'end': 14, 'duration': 4, 'demands': [1]}, {'task': 'T8', 'start': 10, 'end': 14, 'duration': 4, 'demands': [1]}, {'task': 'T9', 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}, {'task': 'T10', 'start': 12, 'end': 14, 'duration': 2, 'demands': [3]}]",15,json,0
RCPSP,RCPSP,"Many projects look simple on paper until the crews and tools start to overlap; this one lists hardscape and planting phases that each take a certain amount of time and need certain crews or equipment while they run. The decision is to place each phase on the timeline, respecting which phases must finish before others can begin, and avoiding any overlap that would demand more resources than exist. A smarter placement is the one where the final phase finishes sooner — that finish time is the bottom-line measure — and the concrete task breakdown and resource numbers are shown below.

{
  ""total_phases"": 10,
  ""resource_types_count"": 3,
  ""resources"": [
    1,
    2,
    3
  ],
  ""available_resources"": [
    {
      ""resource_id"": 1,
      ""capacity"": 3
    },
    {
      ""resource_id"": 2,
      ""capacity"": 2
    },
    {
      ""resource_id"": 3,
      ""capacity"": 3
    }
  ],
  ""tasks"": [
    {
      ""phase_id"": ""T1"",
      ""phase_duration"": 1,
      ""resource_requirements"": [
        0,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T2"",
        ""T3"",
        ""T4""
      ]
    },
    {
      ""phase_id"": ""T2"",
      ""phase_duration"": 7,
      ""resource_requirements"": [
        2,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T7""
      ]
    },
    {
      ""phase_id"": ""T3"",
      ""phase_duration"": 4,
      ""resource_requirements"": [
        0,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T5""
      ]
    },
    {
      ""phase_id"": ""T4"",
      ""phase_duration"": 3,
      ""resource_requirements"": [
        0,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T8""
      ]
    },
    {
      ""phase_id"": ""T5"",
      ""phase_duration"": 3,
      ""resource_requirements"": [
        2,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T8""
      ]
    },
    {
      ""phase_id"": ""T6"",
      ""phase_duration"": 1,
      ""resource_requirements"": [
        0,
        0,
        0
      ],
      ""finish_to_start_successors"": [
        ""T10""
      ]
    },
    {
      ""phase_id"": ""T7"",
      ""phase_duration"": 2,
      ""resource_requirements"": [
        0,
        0,
        1
      ],
      ""finish_to_start_successors"": [
        ""T9""
      ]
    },
    {
      ""phase_id"": ""T8"",
      ""phase_duration"": 3,
      ""resource_requirements"": [
        0,
        0,
        2
      ],
      ""finish_to_start_successors"": [
        ""T9""
      ]
    },
    {
      ""phase_id"": ""T9"",
      ""phase_duration"": 4,
      ""resource_requirements"": [
        0,
        1,
        0
      ],
      ""finish_to_start_successors"": []
    },
    {
      ""phase_id"": ""T10"",
      ""phase_duration"": 5,
      ""resource_requirements"": [
        0,
        0,
        0
      ],
      ""finish_to_start_successors"": []
    }
  ]
}

Also, when you send the actual schedule back, please use this simple JSON shape — nothing fancy, just a list of tasks with their start and end times in that structure:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little form: each entry names the task and the time window when it runs. This block is just the sketch of the shape I need — replace those placeholders with the real task identifiers and numbers for start and end times when you give the schedule.

Please make sure to use the exact task identifiers from the instance input — do not rename tasks or invent new labels.
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""

That's it — fill in that JSON shape with the real schedule and we'll be set.","{'nr_tasks': 10, 'nr_resources': 3, 'capacities': [3, 2, 3], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2, 3]}, {'duration': 7, 'demands': [2, 0, 0], 'successors': [6]}, {'duration': 4, 'demands': [0, 0, 0], 'successors': [4]}, {'duration': 3, 'demands': [0, 0, 0], 'successors': [7]}, {'duration': 3, 'demands': [2, 0, 0], 'successors': [7]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': [9]}, {'duration': 2, 'demands': [0, 0, 1], 'successors': [8]}, {'duration': 3, 'demands': [0, 0, 2], 'successors': [8]}, {'duration': 4, 'demands': [0, 1, 0], 'successors': []}, {'duration': 5, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 1, 'end': 8, 'duration': 7, 'demands': [2, 0, 0]}, {'task': 2, 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 3, 'start': 5, 'end': 8, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 4, 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 0, 0]}, {'task': 5, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 6, 'start': 9, 'end': 11, 'duration': 2, 'demands': [0, 0, 1]}, {'task': 7, 'start': 11, 'end': 14, 'duration': 3, 'demands': [0, 0, 2]}, {'task': 8, 'start': 14, 'end': 18, 'duration': 4, 'demands': [0, 1, 0]}, {'task': 9, 'start': 10, 'end': 15, 'duration': 5, 'demands': [0, 0, 0]}]",18.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [1, 2, 3], 'resource_capacities': [{'resource_id': 1, 'capacity': 3}, {'resource_id': 2, 'capacity': 2}, {'resource_id': 3, 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3', 'T4']}, {'task_id': 'T2', 'duration': 7, 'demands': [2, 0, 0], 'successors': ['T7']}, {'task_id': 'T3', 'duration': 4, 'demands': [0, 0, 0], 'successors': ['T5']}, {'task_id': 'T4', 'duration': 3, 'demands': [0, 0, 0], 'successors': ['T8']}, {'task_id': 'T5', 'duration': 3, 'demands': [2, 0, 0], 'successors': ['T8']}, {'task_id': 'T6', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T10']}, {'task_id': 'T7', 'duration': 2, 'demands': [0, 0, 1], 'successors': ['T9']}, {'task_id': 'T8', 'duration': 3, 'demands': [0, 0, 2], 'successors': ['T9']}, {'task_id': 'T9', 'duration': 4, 'demands': [0, 1, 0], 'successors': []}, {'task_id': 'T10', 'duration': 5, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 1, 'end': 8, 'duration': 7, 'demands': [2, 0, 0]}, {'task': 'T3', 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 'T4', 'start': 5, 'end': 8, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 'T5', 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 0, 0]}, {'task': 'T6', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T7', 'start': 9, 'end': 11, 'duration': 2, 'demands': [0, 0, 1]}, {'task': 'T8', 'start': 11, 'end': 14, 'duration': 3, 'demands': [0, 0, 2]}, {'task': 'T9', 'start': 14, 'end': 18, 'duration': 4, 'demands': [0, 1, 0]}, {'task': 'T10', 'start': 10, 'end': 15, 'duration': 5, 'demands': [0, 0, 0]}]",16,json,1
RCPSP,RCPSP,"Imagine the behind-the-scenes scramble: every booth and performance block has a known setup length, needs certain setup teams and power lines while it’s being built, and a few of them have to wait for others to be done first. The organizer’s job is to place each job on the timeline with a start time so overlapping work never asks for more teams or electricity than exists, nothing gets omitted or duplicated, and the full site is ready at the earliest moment — basically shave down the time when the last build completes. The specifics for each job and the available crews and power are below.

There are 9 stalls and performance blocks to schedule, 3 resource types to track, and resource capacities are .
For task T1, the organizer must place a 2-time activity that consumes 3, 4, 2 while running and cannot start until T2, T5, T6 have finished.
For task T2, the organizer must place a 2-time activity that consumes 6, 4, 1 while running and cannot start until T3 have finished.
For task T3, the organizer must place a 5-time activity that consumes 3, 0, 2 while running and cannot start until T7 have finished.
For task T4, the organizer must place a 6-time activity that consumes 2, 1, 1 while running and cannot start until T5, T2, T6 have finished.
For task T5, the organizer must place a 3-time activity that consumes 4, 1, 3 while running and cannot start until T9, T8 have finished.
For task T6, the organizer must place a 6-time activity that consumes 2, 3, 5 while running and cannot start until T7, T9 have finished.
For task T7, the organizer must place a 2-time activity that consumes 8, 2, 0 while running and cannot start until T8 have finished.
For task T8, the organizer must place a 2-time activity that consumes 4, 5, 2 while running and cannot start until T9 have finished.
For task T9, the organizer must place a 1-time activity that consumes 0, 0, 0 while running and cannot start until none have finished.
The organizer will use these lines to schedule each job and shave down the time when the last build completes.

Also, when you hand me the schedule, please use this simple JSON shape so it's quick to parse and nobody has to guess what goes where.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just means the top-level ""solution"" is a list of entries; each entry names a task and the time it starts and finishes. Think of ""task"" as the job label (which matches whatever label appears in the input), and ""start"" / ""end"" as the scheduled times — super straightforward, like filling in a short form. This block is only a sketch of the expected shape, not the actual schedule.

Please make sure you use the exact identifiers from the instance input — don't rename them or invent new labels.
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 3, 'capacities': [8, 9, 6], 'tasks': [{'duration': 2, 'demands': [3, 4, 2], 'successors': [1, 4, 5]}, {'duration': 2, 'demands': [6, 4, 1], 'successors': [2]}, {'duration': 5, 'demands': [3, 0, 2], 'successors': [6]}, {'duration': 6, 'demands': [2, 1, 1], 'successors': [4, 1, 5]}, {'duration': 3, 'demands': [4, 1, 3], 'successors': [8, 7]}, {'duration': 6, 'demands': [2, 3, 5], 'successors': [6, 8]}, {'duration': 2, 'demands': [8, 2, 0], 'successors': [7]}, {'duration': 2, 'demands': [4, 5, 2], 'successors': [8]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [3, 4, 2]}, {'task': 1, 'start': 6, 'end': 8, 'duration': 2, 'demands': [6, 4, 1]}, {'task': 2, 'start': 12, 'end': 17, 'duration': 5, 'demands': [3, 0, 2]}, {'task': 3, 'start': 0, 'end': 6, 'duration': 6, 'demands': [2, 1, 1]}, {'task': 4, 'start': 12, 'end': 15, 'duration': 3, 'demands': [4, 1, 3]}, {'task': 5, 'start': 6, 'end': 12, 'duration': 6, 'demands': [2, 3, 5]}, {'task': 6, 'start': 17, 'end': 19, 'duration': 2, 'demands': [8, 2, 0]}, {'task': 7, 'start': 19, 'end': 21, 'duration': 2, 'demands': [4, 5, 2]}, {'task': 8, 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}]",22.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 8}, {'resource_id': 'B', 'capacity': 9}, {'resource_id': 'C', 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [3, 4, 2], 'successors': ['T2', 'T5', 'T6']}, {'task_id': 'T2', 'duration': 2, 'demands': [6, 4, 1], 'successors': ['T3']}, {'task_id': 'T3', 'duration': 5, 'demands': [3, 0, 2], 'successors': ['T7']}, {'task_id': 'T4', 'duration': 6, 'demands': [2, 1, 1], 'successors': ['T5', 'T2', 'T6']}, {'task_id': 'T5', 'duration': 3, 'demands': [4, 1, 3], 'successors': ['T9', 'T8']}, {'task_id': 'T6', 'duration': 6, 'demands': [2, 3, 5], 'successors': ['T7', 'T9']}, {'task_id': 'T7', 'duration': 2, 'demands': [8, 2, 0], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 2, 'demands': [4, 5, 2], 'successors': ['T9']}, {'task_id': 'T9', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [3, 4, 2]}, {'task': 'T2', 'start': 6, 'end': 8, 'duration': 2, 'demands': [6, 4, 1]}, {'task': 'T3', 'start': 12, 'end': 17, 'duration': 5, 'demands': [3, 0, 2]}, {'task': 'T4', 'start': 0, 'end': 6, 'duration': 6, 'demands': [2, 1, 1]}, {'task': 'T5', 'start': 12, 'end': 15, 'duration': 3, 'demands': [4, 1, 3]}, {'task': 'T6', 'start': 6, 'end': 12, 'duration': 6, 'demands': [2, 3, 5]}, {'task': 'T7', 'start': 17, 'end': 19, 'duration': 2, 'demands': [8, 2, 0]}, {'task': 'T8', 'start': 19, 'end': 21, 'duration': 2, 'demands': [4, 5, 2]}, {'task': 'T9', 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}]",17,nl,names
RCPSP,RCPSP,"There’s a remodel to run and the calendar needs filling: every task has a set duration, consumes a few specialists and tools while it runs, and depends on certain earlier jobs being done first. The job is to place each task on the timeline so nothing contradicts those orderings and the crew and kit aren’t double-booked on any day. The winning schedule is the one that finishes the entire renovation sooner — you judge that by when the last task finishes. The exact jobs, durations, and resource counts are listed below.

# total_tasks=6
# resource_types_count=2
# resources=A, B
resource_id,capacity
A,6
B,4
task_identifier,duration_days,resource_demand_vector,immediate_successors
T1,1,""0, 2"",""T2, T4, T3""
T2,6,""2, 2"",""T3, T4, T6""
T3,2,""4, 0"",""T4, T5, T6""
T4,3,""6, 0"",""T6, T5""
T5,3,""1, 1"",""T6""
T6,1,""0, 0"",""none""

Also, when you send the schedule back, please use this simple JSON layout so it's easy to read and check. Here's the shape I expect:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a quick sketch of the shape I need — ""solution"" is a list of task placements. Each item is one task: ""task"" is the task identifier from the instance, ""start"" is when that task begins, and ""end"" is when it finishes. Think of it like filling out a little form for each job: which job, when it starts, and when it ends.

Please note this JSON is only the expected shape, not the actual answer. Also be sure to use the exact identifiers from the input — no renaming and no new labels. 

- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 6, 'nr_resources': 2, 'capacities': [6, 4], 'tasks': [{'duration': 1, 'demands': [0, 2], 'successors': [1, 3, 2]}, {'duration': 6, 'demands': [2, 2], 'successors': [2, 3, 5]}, {'duration': 2, 'demands': [4, 0], 'successors': [3, 4, 5]}, {'duration': 3, 'demands': [6, 0], 'successors': [5, 4]}, {'duration': 3, 'demands': [1, 1], 'successors': [5]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 2]}, {'task': 1, 'start': 1, 'end': 7, 'duration': 6, 'demands': [2, 2]}, {'task': 2, 'start': 7, 'end': 9, 'duration': 2, 'demands': [4, 0]}, {'task': 3, 'start': 9, 'end': 12, 'duration': 3, 'demands': [6, 0]}, {'task': 4, 'start': 12, 'end': 15, 'duration': 3, 'demands': [1, 1]}, {'task': 5, 'start': 15, 'end': 16, 'duration': 1, 'demands': [0, 0]}]",16.0,"{'problem_type': 'RCPSP', 'nr_tasks': 6, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6'], 'resources': ['A', 'B'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 6}, {'resource_id': 'B', 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 2], 'successors': ['T2', 'T4', 'T3']}, {'task_id': 'T2', 'duration': 6, 'demands': [2, 2], 'successors': ['T3', 'T4', 'T6']}, {'task_id': 'T3', 'duration': 2, 'demands': [4, 0], 'successors': ['T4', 'T5', 'T6']}, {'task_id': 'T4', 'duration': 3, 'demands': [6, 0], 'successors': ['T6', 'T5']}, {'task_id': 'T5', 'duration': 3, 'demands': [1, 1], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 2]}, {'task': 'T2', 'start': 1, 'end': 7, 'duration': 6, 'demands': [2, 2]}, {'task': 'T3', 'start': 7, 'end': 9, 'duration': 2, 'demands': [4, 0]}, {'task': 'T4', 'start': 9, 'end': 12, 'duration': 3, 'demands': [6, 0]}, {'task': 'T5', 'start': 12, 'end': 15, 'duration': 3, 'demands': [1, 1]}, {'task': 'T6', 'start': 15, 'end': 16, 'duration': 1, 'demands': [0, 0]}]",18,csv,names
RCPSP,RCPSP,"There’s a relief operation to organize where every distribution job has a fixed duration, consumes some number of trucks and volunteers while it runs, and depends on certain earlier jobs finishing before it can begin. The coordinator must pick a start time for each job, ensuring each one happens once, that simultaneous jobs don’t use more vehicles or people than available, and that the whole operation’s completion time — the moment the last job ends — is as early as possible. The exact activities, timings, and available trucks and volunteers are listed below.

# total_distribution_activities=5
# resource_type_count=3
# resources=0, 1, 2
resource_id,capacity
0,7
1,6
2,5
distribution_job_id,job_duration,resource_demands,dependent_jobs
T1,2,""4, 1, 2"",""T2, T3, T5, T4""
T2,3,""6, 3, 5"",""T5, T4""
T3,2,""3, 2, 2"",""T4, T2, T5""
T4,3,""2, 1, 0"",""T5""
T5,1,""0, 0, 0"",""none""

Also, when you send the schedule back, please use this simple JSON layout so it's easy to read and parse — just a quick sketch of the shape I need:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This shows a list called ""solution"" made of entries for each job: ""task"" is the job identifier, ""start"" is when that job begins, and ""end"" is when it finishes. Think of it like filling out a little form for each activity: which job, and its start and end times. It's just the shape I expect — not the actual schedule itself.

Please make sure to use the exact identifiers from the instance input — do not rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""

I'll check the schedule against those exact IDs, so keeping them identical is important.","{'nr_tasks': 5, 'nr_resources': 3, 'capacities': [7, 6, 5], 'tasks': [{'duration': 2, 'demands': [4, 1, 2], 'successors': [1, 2, 4, 3]}, {'duration': 3, 'demands': [6, 3, 5], 'successors': [4, 3]}, {'duration': 2, 'demands': [3, 2, 2], 'successors': [3, 1, 4]}, {'duration': 3, 'demands': [2, 1, 0], 'successors': [4]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [4, 1, 2]}, {'task': 1, 'start': 4, 'end': 7, 'duration': 3, 'demands': [6, 3, 5]}, {'task': 2, 'start': 2, 'end': 4, 'duration': 2, 'demands': [3, 2, 2]}, {'task': 3, 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 4, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}]",11.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 7}, {'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [4, 1, 2], 'successors': ['T2', 'T3', 'T5', 'T4']}, {'task_id': 'T2', 'duration': 3, 'demands': [6, 3, 5], 'successors': ['T5', 'T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [3, 2, 2], 'successors': ['T4', 'T2', 'T5']}, {'task_id': 'T4', 'duration': 3, 'demands': [2, 1, 0], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [4, 1, 2]}, {'task': 'T2', 'start': 4, 'end': 7, 'duration': 3, 'demands': [6, 3, 5]}, {'task': 'T3', 'start': 2, 'end': 4, 'duration': 2, 'demands': [3, 2, 2]}, {'task': 'T4', 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 'T5', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}]",19,csv,0
RCPSP,RCPSP,"We’re the caterers for a large reception and have to map out when every cooking and plating task happens; each task has a fixed duration and needs certain kitchen gear and people. The plan needs to respect the natural order (some tasks can’t start until others are done), include every required task exactly once, and keep total equipment and staff use within what’s on hand. The way to tell if the plan is better is simple: look at the time the final dish is ready — that time should be as small as possible. Full, concrete task and resource details follow below.

We'll schedule 10 cooking and plating steps across 3 resource types with available quantities .
Step T1: we will run it for 1 minutes; it requires 0, 0, 0 of the resource types while running, and its direct successors are T2, T3.
Step T2: we will run it for 2 minutes; it requires 1, 2, 1 of the resource types while running, and its direct successors are T4, T5, T7.
Step T3: we will run it for 2 minutes; it requires 0, 2, 1 of the resource types while running, and its direct successors are T6.
Step T4: we will run it for 3 minutes; it requires 2, 2, 2 of the resource types while running, and its direct successors are T7, T3.
Step T5: we will run it for 5 minutes; it requires 1, 1, 6 of the resource types while running, and its direct successors are T6.
Step T6: we will run it for 3 minutes; it requires 2, 1, 0 of the resource types while running, and its direct successors are T7.
Step T7: we will run it for 1 minutes; it requires 0, 0, 0 of the resource types while running, and its direct successors are none.
Step T8: we will run it for 2 minutes; it requires 1, 1, 5 of the resource types while running, and its direct successors are T5, T2.
Step T9: we will run it for 4 minutes; it requires 0, 0, 0 of the resource types while running, and its direct successors are T8.
Step T10: we will run it for 1 minutes; it requires 1, 2, 1 of the resource types while running, and its direct successors are none.
We'll respect the listed  and precedence constraints and aim to have the final dish ready as soon as possible.

Also, when you hand me the actual schedule, please put it in this simple JSON shape so it's easy for everyone (and any tools) to read:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little form: ""solution"" is the list of all the tasks; each entry says which task it is and the time it starts and finishes (so you can see who's cooking what when). It's just a sketch of the shape I need here, not the real answer itself.

Please also make sure to use the exact task identifiers from the instance — don't rename them and don't invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 10, 'nr_resources': 3, 'capacities': [6, 5, 6], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2]}, {'duration': 2, 'demands': [1, 2, 1], 'successors': [3, 4, 6]}, {'duration': 2, 'demands': [0, 2, 1], 'successors': [5]}, {'duration': 3, 'demands': [2, 2, 2], 'successors': [6, 2]}, {'duration': 5, 'demands': [1, 1, 6], 'successors': [5]}, {'duration': 3, 'demands': [2, 1, 0], 'successors': [6]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 2, 'demands': [1, 1, 5], 'successors': [4, 1]}, {'duration': 4, 'demands': [0, 0, 0], 'successors': [7]}, {'duration': 1, 'demands': [1, 2, 1], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 6, 'end': 8, 'duration': 2, 'demands': [1, 2, 1]}, {'task': 2, 'start': 16, 'end': 18, 'duration': 2, 'demands': [0, 2, 1]}, {'task': 3, 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 2, 2]}, {'task': 4, 'start': 11, 'end': 16, 'duration': 5, 'demands': [1, 1, 6]}, {'task': 5, 'start': 18, 'end': 21, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 6, 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 4, 'end': 6, 'duration': 2, 'demands': [1, 1, 5]}, {'task': 8, 'start': 0, 'end': 4, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 9, 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 2, 1]}]",22.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 6}, {'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3']}, {'task_id': 'T2', 'duration': 2, 'demands': [1, 2, 1], 'successors': ['T4', 'T5', 'T7']}, {'task_id': 'T3', 'duration': 2, 'demands': [0, 2, 1], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 3, 'demands': [2, 2, 2], 'successors': ['T7', 'T3']}, {'task_id': 'T5', 'duration': 5, 'demands': [1, 1, 6], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 3, 'demands': [2, 1, 0], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 2, 'demands': [1, 1, 5], 'successors': ['T5', 'T2']}, {'task_id': 'T9', 'duration': 4, 'demands': [0, 0, 0], 'successors': ['T8']}, {'task_id': 'T10', 'duration': 1, 'demands': [1, 2, 1], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 6, 'end': 8, 'duration': 2, 'demands': [1, 2, 1]}, {'task': 'T3', 'start': 16, 'end': 18, 'duration': 2, 'demands': [0, 2, 1]}, {'task': 'T4', 'start': 8, 'end': 11, 'duration': 3, 'demands': [2, 2, 2]}, {'task': 'T5', 'start': 11, 'end': 16, 'duration': 5, 'demands': [1, 1, 6]}, {'task': 'T6', 'start': 18, 'end': 21, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 'T7', 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 4, 'end': 6, 'duration': 2, 'demands': [1, 1, 5]}, {'task': 'T9', 'start': 0, 'end': 4, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 'T10', 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 2, 1]}]",20,nl,0
RCPSP,RCPSP,"We’ve got a day full of moves to coordinate: a roster of loading tasks and hauls, each with a set duration, and a few that have to wait for earlier tasks to be completed. The choice is about the start time for each task so no more trucks or crews are used at the same moment than are available, nothing is skipped or duplicated, and every dependency is respected. A better plan is simply the one that gets the final unload done earliest — measure success by the moment the last task ends. The exact tasks, durations, and resource numbers appear below.

{
  ""total_jobs"": 4,
  ""resource_types_count"": 3,
  ""resources"": [
    0,
    1,
    2
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 5
    },
    {
      ""resource_id"": 1,
      ""capacity"": 5
    },
    {
      ""resource_id"": 2,
      ""capacity"": 3
    }
  ],
  ""tasks"": [
    {
      ""job_id"": ""T1"",
      ""job_duration"": 2,
      ""resource_demands"": [
        2,
        3,
        1
      ],
      ""successor_job_ids"": [
        ""T2"",
        ""T3"",
        ""T4""
      ]
    },
    {
      ""job_id"": ""T2"",
      ""job_duration"": 4,
      ""resource_demands"": [
        1,
        1,
        2
      ],
      ""successor_job_ids"": [
        ""T3"",
        ""T4""
      ]
    },
    {
      ""job_id"": ""T3"",
      ""job_duration"": 3,
      ""resource_demands"": [
        1,
        1,
        0
      ],
      ""successor_job_ids"": [
        ""T4""
      ]
    },
    {
      ""job_id"": ""T4"",
      ""job_duration"": 1,
      ""resource_demands"": [
        0,
        0,
        0
      ],
      ""successor_job_ids"": []
    }
  ]
}

To turn that plan into a machine-friendly reply, just drop your schedule into this simple JSON shape — nothing fancy, just a list of tasks with start and end times. Example sketch:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of each object in ""solution"" as one row on a packing slip: ""task"" is the task identifier, ""start"" is when it begins, and ""end"" is when it finishes. Keep it casual — it's just a quick way for me to read your plan.

This JSON block is only a template showing the expected shape, not the actual schedule. Please use the exact identifiers from the instance input — do not rename tasks and do not invent new labels.

- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 4, 'nr_resources': 3, 'capacities': [5, 5, 3], 'tasks': [{'duration': 2, 'demands': [2, 3, 1], 'successors': [1, 2, 3]}, {'duration': 4, 'demands': [1, 1, 2], 'successors': [2, 3]}, {'duration': 3, 'demands': [1, 1, 0], 'successors': [3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 3, 1]}, {'task': 1, 'start': 2, 'end': 6, 'duration': 4, 'demands': [1, 1, 2]}, {'task': 2, 'start': 6, 'end': 9, 'duration': 3, 'demands': [1, 1, 0]}, {'task': 3, 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 5}, {'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [2, 3, 1], 'successors': ['T2', 'T3', 'T4']}, {'task_id': 'T2', 'duration': 4, 'demands': [1, 1, 2], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 3, 'demands': [1, 1, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 3, 1]}, {'task': 'T2', 'start': 2, 'end': 6, 'duration': 4, 'demands': [1, 1, 2]}, {'task': 'T3', 'start': 6, 'end': 9, 'duration': 3, 'demands': [1, 1, 0]}, {'task': 'T4', 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",21,json,0
RCPSP,RCPSP,"Recently the editorial floor has been a tangle of overlapping jobs: fixed-length editing and layout tasks that each pull on limited proofreaders or a layout station, with clear “do this after that” relationships. The challenge is to assign start times to all tasks so nothing’s missed or repeated, resource usage never exceeds the team or equipment on hand at any time, and the whole project wraps up quickly — the speed is judged by the finish time of the last task on the schedule. The detailed instance is shown below.

{
  ""total_editorial_tasks"": 7,
  ""resource_types_count"": 3,
  ""resources"": [
    1,
    2,
    3
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 1,
      ""capacity"": 7
    },
    {
      ""resource_id"": 2,
      ""capacity"": 10
    },
    {
      ""resource_id"": 3,
      ""capacity"": 8
    }
  ],
  ""tasks"": [
    {
      ""task_identifier"": ""T1"",
      ""task_duration"": 3,
      ""resource_demands"": [
        2,
        3,
        1
      ],
      ""successor_task_ids"": [
        ""T2"",
        ""T4""
      ]
    },
    {
      ""task_identifier"": ""T2"",
      ""task_duration"": 3,
      ""resource_demands"": [
        2,
        2,
        2
      ],
      ""successor_task_ids"": [
        ""T4"",
        ""T3"",
        ""T7""
      ]
    },
    {
      ""task_identifier"": ""T3"",
      ""task_duration"": 9,
      ""resource_demands"": [
        1,
        4,
        2
      ],
      ""successor_task_ids"": [
        ""T7""
      ]
    },
    {
      ""task_identifier"": ""T4"",
      ""task_duration"": 2,
      ""resource_demands"": [
        5,
        4,
        2
      ],
      ""successor_task_ids"": [
        ""T5""
      ]
    },
    {
      ""task_identifier"": ""T5"",
      ""task_duration"": 1,
      ""resource_demands"": [
        2,
        1,
        0
      ],
      ""successor_task_ids"": [
        ""T6""
      ]
    },
    {
      ""task_identifier"": ""T6"",
      ""task_duration"": 6,
      ""resource_demands"": [
        7,
        8,
        1
      ],
      ""successor_task_ids"": [
        ""T7""
      ]
    },
    {
      ""task_identifier"": ""T7"",
      ""task_duration"": 1,
      ""resource_demands"": [
        0,
        0,
        0
      ],
      ""successor_task_ids"": []
    }
  ]
}

Also, when you send back the schedule, please keep it in a simple JSON shape like this — just so whoever's reading it can pick out each task and its times quickly:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a small sketch of the shape I expect: ""solution"" is a list of entries, each entry is one task with its identifier under ""task"" and the chosen start and end times under ""start"" and ""end"". Think of it like filling out a little form for each job — which one it is, when it begins, and when it finishes. The JSON here is only the format example, not the actual schedule.

Please use the exact identifiers from the instance input — don't rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 7, 'nr_resources': 3, 'capacities': [7, 10, 8], 'tasks': [{'duration': 3, 'demands': [2, 3, 1], 'successors': [1, 3]}, {'duration': 3, 'demands': [2, 2, 2], 'successors': [3, 2, 6]}, {'duration': 9, 'demands': [1, 4, 2], 'successors': [6]}, {'duration': 2, 'demands': [5, 4, 2], 'successors': [4]}, {'duration': 1, 'demands': [2, 1, 0], 'successors': [5]}, {'duration': 6, 'demands': [7, 8, 1], 'successors': [6]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 3, 'duration': 3, 'demands': [2, 3, 1]}, {'task': 1, 'start': 3, 'end': 6, 'duration': 3, 'demands': [2, 2, 2]}, {'task': 2, 'start': 6, 'end': 15, 'duration': 9, 'demands': [1, 4, 2]}, {'task': 3, 'start': 6, 'end': 8, 'duration': 2, 'demands': [5, 4, 2]}, {'task': 4, 'start': 8, 'end': 9, 'duration': 1, 'demands': [2, 1, 0]}, {'task': 5, 'start': 15, 'end': 21, 'duration': 6, 'demands': [7, 8, 1]}, {'task': 6, 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}]",22.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [1, 2, 3], 'resource_capacities': [{'resource_id': 1, 'capacity': 7}, {'resource_id': 2, 'capacity': 10}, {'resource_id': 3, 'capacity': 8}], 'task_records': [{'task_id': 'T1', 'duration': 3, 'demands': [2, 3, 1], 'successors': ['T2', 'T4']}, {'task_id': 'T2', 'duration': 3, 'demands': [2, 2, 2], 'successors': ['T4', 'T3', 'T7']}, {'task_id': 'T3', 'duration': 9, 'demands': [1, 4, 2], 'successors': ['T7']}, {'task_id': 'T4', 'duration': 2, 'demands': [5, 4, 2], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [2, 1, 0], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 6, 'demands': [7, 8, 1], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 3, 'duration': 3, 'demands': [2, 3, 1]}, {'task': 'T2', 'start': 3, 'end': 6, 'duration': 3, 'demands': [2, 2, 2]}, {'task': 'T3', 'start': 6, 'end': 15, 'duration': 9, 'demands': [1, 4, 2]}, {'task': 'T4', 'start': 6, 'end': 8, 'duration': 2, 'demands': [5, 4, 2]}, {'task': 'T5', 'start': 8, 'end': 9, 'duration': 1, 'demands': [2, 1, 0]}, {'task': 'T6', 'start': 15, 'end': 21, 'duration': 6, 'demands': [7, 8, 1]}, {'task': 'T7', 'start': 21, 'end': 22, 'duration': 1, 'demands': [0, 0, 0]}]",22,json,1
RCPSP,RCPSP,"I’m trying to pull together a tight schedule for all the fittings: each alteration takes a set time, uses up some of the scarce sewing machines and stylists while it’s happening, and a few sessions depend on earlier adjustments. The choice is about when to start each job so the studio never uses more machines or hands than it has, nothing is skipped or repeated, and the whole lineup is finished sooner rather than later — which you measure by the time the last job finishes. The exact list of fittings, their durations and resource needs is given below.

# total_fittings=6
# resource_types_count=1
# resources=0
resource_id,capacity
0,8
fitting_id,fitting_duration,resource_demands,dependent_fittings
T1,1,""0"",""T2, T5, T4, T6""
T2,2,""3"",""T5, T6, T3""
T3,3,""5"",""T6""
T4,2,""2"",""T6, T2, T3""
T5,2,""1"",""T6, T3""
T6,1,""0"",""none""

I'll drop the final schedule back in a simple JSON layout so it's easy to paste into whatever you're using — something like this:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just sketches the shape I’ll use: ""solution"" is an array of the scheduled fittings, and each item gives the task identifier plus its numeric start and end times. Think of each object as a little form for one fitting: which job it is, when it starts, and when it finishes. That JSON is just the expected shape — I’ll fill in the actual task IDs and numbers for the real schedule.

Please make sure to use the exact task identifiers from the instance input — don’t rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 6, 'nr_resources': 1, 'capacities': [8], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 4, 3, 5]}, {'duration': 2, 'demands': [3], 'successors': [4, 5, 2]}, {'duration': 3, 'demands': [5], 'successors': [5]}, {'duration': 2, 'demands': [2], 'successors': [5, 1, 2]}, {'duration': 2, 'demands': [1], 'successors': [5, 2]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 3, 'end': 5, 'duration': 2, 'demands': [3]}, {'task': 2, 'start': 7, 'end': 10, 'duration': 3, 'demands': [5]}, {'task': 3, 'start': 1, 'end': 3, 'duration': 2, 'demands': [2]}, {'task': 4, 'start': 5, 'end': 7, 'duration': 2, 'demands': [1]}, {'task': 5, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0]}]",11.0,"{'problem_type': 'RCPSP', 'nr_tasks': 6, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6'], 'resources': [0], 'resource_capacities': [{'resource_id': 0, 'capacity': 8}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T5', 'T4', 'T6']}, {'task_id': 'T2', 'duration': 2, 'demands': [3], 'successors': ['T5', 'T6', 'T3']}, {'task_id': 'T3', 'duration': 3, 'demands': [5], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 2, 'demands': [2], 'successors': ['T6', 'T2', 'T3']}, {'task_id': 'T5', 'duration': 2, 'demands': [1], 'successors': ['T6', 'T3']}, {'task_id': 'T6', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 3, 'end': 5, 'duration': 2, 'demands': [3]}, {'task': 'T3', 'start': 7, 'end': 10, 'duration': 3, 'demands': [5]}, {'task': 'T4', 'start': 1, 'end': 3, 'duration': 2, 'demands': [2]}, {'task': 'T5', 'start': 5, 'end': 7, 'duration': 2, 'demands': [1]}, {'task': 'T6', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0]}]",23,csv,0
RCPSP,RCPSP,"I like to imagine the studio as a busy kitchen where each TV segment is a dish that takes a fixed time to prepare. The job is to decide when each dish starts simmering so that the ovens (studios) and chefs (crews) aren’t asked to handle more than they can at once, and pieces that need to be plated after another are started only after their predecessors finish. A better plan is simply the one where the whole menu is ready earlier — that is, the clock time when the last segment prep finishes should be as small as possible. Every prep must get a single start time (no skipping, no doubling up), and at no moment can the number of studios or crew in use exceed what’s available. The concrete details about segment lengths, crew counts and studio limits are shown below.

I’ll list the 8 segment preps that follow, the 3 limited resource types, and the available capacities for them: .

| resource_id | capacity |
|---|---|
| 0 | 5 |
| 1 | 5 |
| 2 | 2 |

| segment_id | prep_time | resource_requirements | depends_on_finished_before_start |
|---|---|---|---|
| T1 | 1 | 0, 0, 0 | T2, T3, T5 |
| T2 | 1 | 4, 3, 1 | T4, T5 |
| T3 | 2 | 0, 2, 1 | T6, T7 |
| T4 | 2 | 5, 3, 2 | T7, T6 |
| T5 | 5 | 1, 2, 2 | T6, T8 |
| T6 | 3 | 1, 1, 0 | T7 |
| T7 | 1 | 0, 0, 0 | none |
| T8 | 4 | 0, 3, 1 | none |

I’ll use these particulars to assign a single start time to each prep so ovens and chefs aren’t overloaded and the whole menu finishes as early as possible.

If you're going to hand the plan back to me, a simple JSON shape works great — nothing fancy, just a list of which segment starts when and when it finishes. Something like this:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little form: ""solution"" is the list of scheduled preps, each item names the task (the segment), the time it starts, and the time it ends. Super informal: list every segment once with its start and end times. This block is just the sketch of the shape I expect, not the actual schedule — you'll fill in the real task IDs and times.

Please use the exact identifiers from the instance input — don't rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 8, 'nr_resources': 3, 'capacities': [5, 5, 2], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2, 4]}, {'duration': 1, 'demands': [4, 3, 1], 'successors': [3, 4]}, {'duration': 2, 'demands': [0, 2, 1], 'successors': [5, 6]}, {'duration': 2, 'demands': [5, 3, 2], 'successors': [6, 5]}, {'duration': 5, 'demands': [1, 2, 2], 'successors': [5, 7]}, {'duration': 3, 'demands': [1, 1, 0], 'successors': [6]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 4, 'demands': [0, 3, 1], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 1, 'end': 2, 'duration': 1, 'demands': [4, 3, 1]}, {'task': 2, 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 2, 1]}, {'task': 3, 'start': 8, 'end': 10, 'duration': 2, 'demands': [5, 3, 2]}, {'task': 4, 'start': 3, 'end': 8, 'duration': 5, 'demands': [1, 2, 2]}, {'task': 5, 'start': 10, 'end': 13, 'duration': 3, 'demands': [1, 1, 0]}, {'task': 6, 'start': 13, 'end': 14, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 10, 'end': 14, 'duration': 4, 'demands': [0, 3, 1]}]",14.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 5}, {'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 2}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3', 'T5']}, {'task_id': 'T2', 'duration': 1, 'demands': [4, 3, 1], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 2, 'demands': [0, 2, 1], 'successors': ['T6', 'T7']}, {'task_id': 'T4', 'duration': 2, 'demands': [5, 3, 2], 'successors': ['T7', 'T6']}, {'task_id': 'T5', 'duration': 5, 'demands': [1, 2, 2], 'successors': ['T6', 'T8']}, {'task_id': 'T6', 'duration': 3, 'demands': [1, 1, 0], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 4, 'demands': [0, 3, 1], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 1, 'end': 2, 'duration': 1, 'demands': [4, 3, 1]}, {'task': 'T3', 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 2, 1]}, {'task': 'T4', 'start': 8, 'end': 10, 'duration': 2, 'demands': [5, 3, 2]}, {'task': 'T5', 'start': 3, 'end': 8, 'duration': 5, 'demands': [1, 2, 2]}, {'task': 'T6', 'start': 10, 'end': 13, 'duration': 3, 'demands': [1, 1, 0]}, {'task': 'T7', 'start': 13, 'end': 14, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 10, 'end': 14, 'duration': 4, 'demands': [0, 3, 1]}]",24,markdown_table,0
RCPSP,RCPSP,"Many mornings there’s more work than benches or hands, so the shop manager has to decide when each diagnostic and repair starts: each step takes a fixed amount of time, ties up limited benches and technicians while it runs, and some fixes can’t start until earlier checks are done and parts are available. The better the plan, the sooner the last appliance is done — to judge that, look at when every job finishes and the latest of those times is the total completion time. Practical constraints are simple: don’t skip or duplicate tasks, never use more benches or technicians than exist at the same time, and always respect the needed order of tasks. The detailed list of tasks and resource limits is below.

# total_jobs=9
# resource_types=2
# resources=A, B
resource_id,capacity
A,5
B,4
job_id,job_duration,resource_demands,following_jobs
T1,1,""0, 0"",""T2, T8, T3""
T2,2,""0, 4"",""T3, T5, T6""
T3,4,""2, 3"",""T4, T6""
T4,3,""2, 0"",""T6, T7, T8""
T5,6,""2, 0"",""T7""
T6,3,""5, 0"",""T8""
T7,4,""1, 4"",""T8, T9""
T8,1,""0, 0"",""none""
T9,3,""2, 0"",""none""

When you hand me a schedule, just follow this simple JSON shape so everything's clear and machine-friendly:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little form: ""solution"" is the list of planned steps, each entry names the task and the time it starts and finishes. It's just a sketch of the shape I expect — not the actual schedule yet.

Please make sure to use the exact identifiers from the instance input — don't rename tasks or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 2, 'capacities': [5, 4], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1, 7, 2]}, {'duration': 2, 'demands': [0, 4], 'successors': [2, 4, 5]}, {'duration': 4, 'demands': [2, 3], 'successors': [3, 5]}, {'duration': 3, 'demands': [2, 0], 'successors': [5, 6, 7]}, {'duration': 6, 'demands': [2, 0], 'successors': [6]}, {'duration': 3, 'demands': [5, 0], 'successors': [7]}, {'duration': 4, 'demands': [1, 4], 'successors': [7, 8]}, {'duration': 1, 'demands': [0, 0], 'successors': []}, {'duration': 3, 'demands': [2, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 4]}, {'task': 2, 'start': 3, 'end': 7, 'duration': 4, 'demands': [2, 3]}, {'task': 3, 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 0]}, {'task': 4, 'start': 3, 'end': 9, 'duration': 6, 'demands': [2, 0]}, {'task': 5, 'start': 14, 'end': 17, 'duration': 3, 'demands': [5, 0]}, {'task': 6, 'start': 10, 'end': 14, 'duration': 4, 'demands': [1, 4]}, {'task': 7, 'start': 17, 'end': 18, 'duration': 1, 'demands': [0, 0]}, {'task': 8, 'start': 17, 'end': 20, 'duration': 3, 'demands': [2, 0]}]",20.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': ['A', 'B'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 5}, {'resource_id': 'B', 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2', 'T8', 'T3']}, {'task_id': 'T2', 'duration': 2, 'demands': [0, 4], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 4, 'demands': [2, 3], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 3, 'demands': [2, 0], 'successors': ['T6', 'T7', 'T8']}, {'task_id': 'T5', 'duration': 6, 'demands': [2, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 3, 'demands': [5, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 4, 'demands': [1, 4], 'successors': ['T8', 'T9']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}, {'task_id': 'T9', 'duration': 3, 'demands': [2, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [0, 4]}, {'task': 'T3', 'start': 3, 'end': 7, 'duration': 4, 'demands': [2, 3]}, {'task': 'T4', 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 0]}, {'task': 'T5', 'start': 3, 'end': 9, 'duration': 6, 'demands': [2, 0]}, {'task': 'T6', 'start': 14, 'end': 17, 'duration': 3, 'demands': [5, 0]}, {'task': 'T7', 'start': 10, 'end': 14, 'duration': 4, 'demands': [1, 4]}, {'task': 'T8', 'start': 17, 'end': 18, 'duration': 1, 'demands': [0, 0]}, {'task': 'T9', 'start': 17, 'end': 20, 'duration': 3, 'demands': [2, 0]}]",25,csv,names
RCPSP,RCPSP,"Recently the maintenance team faced a backlog: a list of fixed-length jobs for different aircraft, each occupying a bay and some techs while it’s on the clock, and a few jobs that must wait for inspections first. The task was to schedule start times for all jobs so nothing gets left out or duplicated, required orderings are kept, and the limited hangars and technicians aren’t over-committed at any time. The aim is to have the fleet ready again as soon as possible — judged by the time the last job finishes. The specific jobs, durations, and resource limits are shown below.

{
  ""total_number_of_jobs"": 8,
  ""number_of_resource_types"": 1,
  ""resources"": [
    1
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 1,
      ""capacity"": 3
    }
  ],
  ""tasks"": [
    {
      ""job_identifier"": ""T1"",
      ""job_duration"": 1,
      ""resource_demands"": [
        0
      ],
      ""immediate_successor_jobs"": [
        ""T2"",
        ""T7""
      ]
    },
    {
      ""job_identifier"": ""T2"",
      ""job_duration"": 4,
      ""resource_demands"": [
        3
      ],
      ""immediate_successor_jobs"": [
        ""T3"",
        ""T5""
      ]
    },
    {
      ""job_identifier"": ""T3"",
      ""job_duration"": 1,
      ""resource_demands"": [
        3
      ],
      ""immediate_successor_jobs"": [
        ""T4""
      ]
    },
    {
      ""job_identifier"": ""T4"",
      ""job_duration"": 3,
      ""resource_demands"": [
        3
      ],
      ""immediate_successor_jobs"": [
        ""T8""
      ]
    },
    {
      ""job_identifier"": ""T5"",
      ""job_duration"": 3,
      ""resource_demands"": [
        3
      ],
      ""immediate_successor_jobs"": [
        ""T6""
      ]
    },
    {
      ""job_identifier"": ""T6"",
      ""job_duration"": 4,
      ""resource_demands"": [
        3
      ],
      ""immediate_successor_jobs"": [
        ""T8""
      ]
    },
    {
      ""job_identifier"": ""T7"",
      ""job_duration"": 2,
      ""resource_demands"": [
        1
      ],
      ""immediate_successor_jobs"": [
        ""T8""
      ]
    },
    {
      ""job_identifier"": ""T8"",
      ""job_duration"": 1,
      ""resource_demands"": [
        0
      ],
      ""immediate_successor_jobs"": []
    }
  ]
}

When you send the schedule back, just put it in this little JSON shape so it's easy to read and load:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that like a simple form: ""solution"" is a list of all the jobs you're scheduling, and each entry names the task and the start and end times you picked for it. Super informal — it's just a sketch of the shape I need, not the real schedule.

Please make sure to use the exact identifiers from the instance input — don't rename tasks or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 8, 'nr_resources': 1, 'capacities': [3], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 6]}, {'duration': 4, 'demands': [3], 'successors': [2, 4]}, {'duration': 1, 'demands': [3], 'successors': [3]}, {'duration': 3, 'demands': [3], 'successors': [7]}, {'duration': 3, 'demands': [3], 'successors': [5]}, {'duration': 4, 'demands': [3], 'successors': [7]}, {'duration': 2, 'demands': [1], 'successors': [7]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 5, 'duration': 4, 'demands': [3]}, {'task': 2, 'start': 8, 'end': 9, 'duration': 1, 'demands': [3]}, {'task': 3, 'start': 13, 'end': 16, 'duration': 3, 'demands': [3]}, {'task': 4, 'start': 5, 'end': 8, 'duration': 3, 'demands': [3]}, {'task': 5, 'start': 9, 'end': 13, 'duration': 4, 'demands': [3]}, {'task': 6, 'start': 16, 'end': 18, 'duration': 2, 'demands': [1]}, {'task': 7, 'start': 18, 'end': 19, 'duration': 1, 'demands': [0]}]",19.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T7']}, {'task_id': 'T2', 'duration': 4, 'demands': [3], 'successors': ['T3', 'T5']}, {'task_id': 'T3', 'duration': 1, 'demands': [3], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 3, 'demands': [3], 'successors': ['T8']}, {'task_id': 'T5', 'duration': 3, 'demands': [3], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 4, 'demands': [3], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 2, 'demands': [1], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 5, 'duration': 4, 'demands': [3]}, {'task': 'T3', 'start': 8, 'end': 9, 'duration': 1, 'demands': [3]}, {'task': 'T4', 'start': 13, 'end': 16, 'duration': 3, 'demands': [3]}, {'task': 'T5', 'start': 5, 'end': 8, 'duration': 3, 'demands': [3]}, {'task': 'T6', 'start': 9, 'end': 13, 'duration': 4, 'demands': [3]}, {'task': 'T7', 'start': 16, 'end': 18, 'duration': 2, 'demands': [1]}, {'task': 'T8', 'start': 18, 'end': 19, 'duration': 1, 'demands': [0]}]",26,json,1
RCPSP,RCPSP,"Many people on the crew are watching the rehearsal board: each rehearsal block has a fixed length and needs both the stage and certain crew members, and some blocks depend on others finishing first. The challenge was picking exact start times so every block is scheduled exactly once, the order rules are followed, and we never ask for more technicians or stages than we have at any minute. A better lineup is the one where the final end time is sooner — to judge any lineup, find the end times of all blocks, take the latest one, and prefer the lineup with the earlier latest end. The specific rehearsals and resource limits are listed below.

The specific rehearsals and resource limits are listed below: there are 7 rehearsal blocks, 1 resource types, and the resource capacities are .
Block T1 lasts 1 minutes, requires 0 while running, and must finish before T2, T6, T4 start.
Block T2 lasts 6 minutes, requires 5 while running, and must finish before T4 start.
Block T3 lasts 2 minutes, requires 3 while running, and must finish before T7, T2 start.
Block T4 lasts 2 minutes, requires 5 while running, and must finish before T5 start.
Block T5 lasts 4 minutes, requires 3 while running, and must finish before T7 start.
Block T6 lasts 2 minutes, requires 2 while running, and must finish before T7 start.
Block T7 lasts 1 minutes, requires 0 while running, and must finish before none start.
We will judge lineups by the latest block end and prefer the lineup with the earlier latest end.

Also, when you send your lineup back, please stick to a simple JSON shape so it's easy to read and check. Something like this is what I expect:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a sketch of the shape: ""solution"" is a list of rehearsal blocks, and each block entry names the block under ""task"" and the chosen ""start"" and ""end"" times for it. Keep it casual — start and end are just the times on the rehearsal clock (use the same units the instance uses).

The JSON above is only the expected shape, not the actual answer. Please use the exact identifiers from the instance input — do not rename or invent labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 7, 'nr_resources': 1, 'capacities': [5], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 5, 3]}, {'duration': 6, 'demands': [5], 'successors': [3]}, {'duration': 2, 'demands': [3], 'successors': [6, 1]}, {'duration': 2, 'demands': [5], 'successors': [4]}, {'duration': 4, 'demands': [3], 'successors': [6]}, {'duration': 2, 'demands': [2], 'successors': [6]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 2, 'end': 8, 'duration': 6, 'demands': [5]}, {'task': 2, 'start': 0, 'end': 2, 'duration': 2, 'demands': [3]}, {'task': 3, 'start': 8, 'end': 10, 'duration': 2, 'demands': [5]}, {'task': 4, 'start': 10, 'end': 14, 'duration': 4, 'demands': [3]}, {'task': 5, 'start': 10, 'end': 12, 'duration': 2, 'demands': [2]}, {'task': 6, 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}]",15.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': ['A'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T6', 'T4']}, {'task_id': 'T2', 'duration': 6, 'demands': [5], 'successors': ['T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [3], 'successors': ['T7', 'T2']}, {'task_id': 'T4', 'duration': 2, 'demands': [5], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 4, 'demands': [3], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 2, 'demands': [2], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 2, 'end': 8, 'duration': 6, 'demands': [5]}, {'task': 'T3', 'start': 0, 'end': 2, 'duration': 2, 'demands': [3]}, {'task': 'T4', 'start': 8, 'end': 10, 'duration': 2, 'demands': [5]}, {'task': 'T5', 'start': 10, 'end': 14, 'duration': 4, 'demands': [3]}, {'task': 'T6', 'start': 10, 'end': 12, 'duration': 2, 'demands': [2]}, {'task': 'T7', 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}]",27,nl,names
RCPSP,RCPSP,"Someone has to coordinate the roadworks so traffic can flow again: each repair takes a fixed block of time, consumes some crew hours and a machine while it’s underway, and a few jobs can’t begin until others are done. The job is to pick start times for each activity so the final reopening happens at the earliest possible moment — compare plans by the time the last task finishes, and earlier wins. At no point can the total crews or machines in use exceed what’s on the books, and every repair must be carried out once (no skipping or double-booking). The specific instance details are shown below.

# num_repairs=8
# num_resource_types=3
# resources=0, 1, 2
resource_id,capacity
0,2
1,2
2,2
repair_id,repair_duration,resource_demands,dependent_repairs
T1,4,""0, 0, 0"",""T3, T4""
T2,3,""0, 0, 0"",""T4, T6""
T3,3,""2, 0, 0"",""T4, T6""
T4,2,""0, 0, 0"",""T8""
T5,3,""0, 0, 1"",""T7, T2, T1""
T6,1,""0, 0, 2"",""T7""
T7,4,""0, 1, 0"",""T8""
T8,9,""0, 0, 0"",""none""

Also, when you hand back a plan, please use a simple JSON layout so it's easy to double-check automatically. Something like this is what I expect to see:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This little sketch means: ""solution"" is a list of scheduled activities; each item shows which task it is (task), when it begins (start) and when it finishes (end). Think of it like filling out a short form for every repair — one line per job with its start and end times. The JSON above is just the expected shape, not the actual schedule.

Please be sure to use the exact task identifiers from the instance input — don't rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 8, 'nr_resources': 3, 'capacities': [2, 2, 2], 'tasks': [{'duration': 4, 'demands': [0, 0, 0], 'successors': [2, 3]}, {'duration': 3, 'demands': [0, 0, 0], 'successors': [3, 5]}, {'duration': 3, 'demands': [2, 0, 0], 'successors': [3, 5]}, {'duration': 2, 'demands': [0, 0, 0], 'successors': [7]}, {'duration': 3, 'demands': [0, 0, 1], 'successors': [6, 1, 0]}, {'duration': 1, 'demands': [0, 0, 2], 'successors': [6]}, {'duration': 4, 'demands': [0, 1, 0], 'successors': [7]}, {'duration': 9, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 3, 'end': 7, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 1, 'start': 7, 'end': 10, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 2, 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 0, 0]}, {'task': 3, 'start': 13, 'end': 15, 'duration': 2, 'demands': [0, 0, 0]}, {'task': 4, 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 0, 1]}, {'task': 5, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 2]}, {'task': 6, 'start': 11, 'end': 15, 'duration': 4, 'demands': [0, 1, 0]}, {'task': 7, 'start': 15, 'end': 24, 'duration': 9, 'demands': [0, 0, 0]}]",24.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 2}, {'resource_id': 1, 'capacity': 2}, {'resource_id': 2, 'capacity': 2}], 'task_records': [{'task_id': 'T1', 'duration': 4, 'demands': [0, 0, 0], 'successors': ['T3', 'T4']}, {'task_id': 'T2', 'duration': 3, 'demands': [0, 0, 0], 'successors': ['T4', 'T6']}, {'task_id': 'T3', 'duration': 3, 'demands': [2, 0, 0], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 2, 'demands': [0, 0, 0], 'successors': ['T8']}, {'task_id': 'T5', 'duration': 3, 'demands': [0, 0, 1], 'successors': ['T7', 'T2', 'T1']}, {'task_id': 'T6', 'duration': 1, 'demands': [0, 0, 2], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 4, 'demands': [0, 1, 0], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 9, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 3, 'end': 7, 'duration': 4, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 7, 'end': 10, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 'T3', 'start': 7, 'end': 10, 'duration': 3, 'demands': [2, 0, 0]}, {'task': 'T4', 'start': 13, 'end': 15, 'duration': 2, 'demands': [0, 0, 0]}, {'task': 'T5', 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 0, 1]}, {'task': 'T6', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 2]}, {'task': 'T7', 'start': 11, 'end': 15, 'duration': 4, 'demands': [0, 1, 0]}, {'task': 'T8', 'start': 15, 'end': 24, 'duration': 9, 'demands': [0, 0, 0]}]",28,csv,0
RCPSP,RCPSP,"We’re prepping the shop for opening: there’s a bunch of fixed-length jobs — unloading pallets, unpacking products, setting up displays — and each one ties up forklifts and staff while it runs. Some steps have to wait for others (you can’t set up a display until the items are unpacked), so the decision is when to start each job so nothing overlaps the available forklifts or people and nothing gets skipped or done twice. A better plan gets everything done sooner, measured by the moment the final job wraps up. The specific tasks and capacities are listed below.

# total_tasks=10
# total_resource_types=2
# resources=0, 1
resource_id,capacity
0,3
1,4
task_identifier,duration_minutes,resource_demands,immediate_successors
T1,1,""0, 0"",""T2""
T2,1,""0, 2"",""T3, T5, T6""
T3,3,""1, 3"",""T4, T6""
T4,3,""3, 0"",""T6, T7""
T5,15,""1, 0"",""T7, T3""
T6,6,""2, 0"",""T8""
T7,4,""1, 4"",""T8""
T8,1,""0, 0"",""T10""
T9,1,""1, 4"",""none""
T10,13,""2, 0"",""none""

Just so we're on the same page, when you send the schedule back, a relaxed JSON layout like this is perfect — a list called ""solution"" where each item names the task and its start and end times.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a sketch of the shape I expect, not the actual schedule. In plain terms: ""solution"" holds a bunch of task entries; each entry's ""task"" is the task identifier, ""start"" is when that job begins, and ""end"" is when it finishes (in the same time units used in the instance). Keep it simple — like filling out a form for each job.

Please use the exact task identifiers from the instance input — do not rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 10, 'nr_resources': 2, 'capacities': [3, 4], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1]}, {'duration': 1, 'demands': [0, 2], 'successors': [2, 4, 5]}, {'duration': 3, 'demands': [1, 3], 'successors': [3, 5]}, {'duration': 3, 'demands': [3, 0], 'successors': [5, 6]}, {'duration': 15, 'demands': [1, 0], 'successors': [6, 2]}, {'duration': 6, 'demands': [2, 0], 'successors': [7]}, {'duration': 4, 'demands': [1, 4], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': [9]}, {'duration': 1, 'demands': [1, 4], 'successors': []}, {'duration': 13, 'demands': [2, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 2, 'duration': 1, 'demands': [0, 2]}, {'task': 2, 'start': 17, 'end': 20, 'duration': 3, 'demands': [1, 3]}, {'task': 3, 'start': 20, 'end': 23, 'duration': 3, 'demands': [3, 0]}, {'task': 4, 'start': 2, 'end': 17, 'duration': 15, 'demands': [1, 0]}, {'task': 5, 'start': 23, 'end': 29, 'duration': 6, 'demands': [2, 0]}, {'task': 6, 'start': 23, 'end': 27, 'duration': 4, 'demands': [1, 4]}, {'task': 7, 'start': 29, 'end': 30, 'duration': 1, 'demands': [0, 0]}, {'task': 8, 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 4]}, {'task': 9, 'start': 30, 'end': 43, 'duration': 13, 'demands': [2, 0]}]",43.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 3}, {'resource_id': 1, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 1, 'demands': [0, 2], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 3, 'demands': [1, 3], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 3, 'demands': [3, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 15, 'demands': [1, 0], 'successors': ['T7', 'T3']}, {'task_id': 'T6', 'duration': 6, 'demands': [2, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 4, 'demands': [1, 4], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': ['T10']}, {'task_id': 'T9', 'duration': 1, 'demands': [1, 4], 'successors': []}, {'task_id': 'T10', 'duration': 13, 'demands': [2, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 2, 'duration': 1, 'demands': [0, 2]}, {'task': 'T3', 'start': 17, 'end': 20, 'duration': 3, 'demands': [1, 3]}, {'task': 'T4', 'start': 20, 'end': 23, 'duration': 3, 'demands': [3, 0]}, {'task': 'T5', 'start': 2, 'end': 17, 'duration': 15, 'demands': [1, 0]}, {'task': 'T6', 'start': 23, 'end': 29, 'duration': 6, 'demands': [2, 0]}, {'task': 'T7', 'start': 23, 'end': 27, 'duration': 4, 'demands': [1, 4]}, {'task': 'T8', 'start': 29, 'end': 30, 'duration': 1, 'demands': [0, 0]}, {'task': 'T9', 'start': 0, 'end': 1, 'duration': 1, 'demands': [1, 4]}, {'task': 'T10', 'start': 30, 'end': 43, 'duration': 13, 'demands': [2, 0]}]",29,csv,0
RCPSP,RCPSP,"Many teams are waiting for a schedule: every procedure on the list takes a known amount of time and uses up certain staff and an operating room while it’s happening, and some procedures must wait until others finish. The core decision is to choose start times for all procedures so that the whole set finishes as fast as possible — you tell which schedule is better by checking the time the last procedure ends, and smaller is better. Real-world limits apply: no procedure is dropped or duplicated, precedence pairs must be respected, and at any moment the number of rooms and staff in use can’t exceed capacity. The full data for this instance is shown below.

# total_procedures=8
# resource_types_count=2
# resources=1, 2
resource_id,capacity
1,4
2,3
procedure_id,procedure_duration,resource_demands,successor_ids
T1,1,""0, 0"",""T2""
T2,3,""0, 3"",""T3, T5, T6""
T3,6,""2, 1"",""T4, T6""
T4,4,""3, 0"",""T6, T7""
T5,6,""1, 0"",""T7""
T6,5,""2, 0"",""T8""
T7,4,""1, 3"",""T8""
T8,1,""0, 0"",""none""

Oh, and when you send the schedule back, please put it in this simple JSON shape so it's easy to read and check:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little form: ""solution"" is the list of procedures, each item shows which procedure (""task"") and the time it starts and finishes (""start"" and ""end""). Keep the times in the same units the instance uses. This JSON is just a sketch of the shape I expect — you'll fill in the actual task identifiers and numbers.

Please use the identifiers exactly as they appear in the instance input — do not rename or invent new labels. For example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 8, 'nr_resources': 2, 'capacities': [4, 3], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1]}, {'duration': 3, 'demands': [0, 3], 'successors': [2, 4, 5]}, {'duration': 6, 'demands': [2, 1], 'successors': [3, 5]}, {'duration': 4, 'demands': [3, 0], 'successors': [5, 6]}, {'duration': 6, 'demands': [1, 0], 'successors': [6]}, {'duration': 5, 'demands': [2, 0], 'successors': [7]}, {'duration': 4, 'demands': [1, 3], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 3]}, {'task': 2, 'start': 4, 'end': 10, 'duration': 6, 'demands': [2, 1]}, {'task': 3, 'start': 10, 'end': 14, 'duration': 4, 'demands': [3, 0]}, {'task': 4, 'start': 4, 'end': 10, 'duration': 6, 'demands': [1, 0]}, {'task': 5, 'start': 14, 'end': 19, 'duration': 5, 'demands': [2, 0]}, {'task': 6, 'start': 14, 'end': 18, 'duration': 4, 'demands': [1, 3]}, {'task': 7, 'start': 19, 'end': 20, 'duration': 1, 'demands': [0, 0]}]",20.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 4}, {'resource_id': 2, 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 3, 'demands': [0, 3], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 6, 'demands': [2, 1], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 4, 'demands': [3, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 6, 'demands': [1, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 5, 'demands': [2, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 4, 'demands': [1, 3], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 3]}, {'task': 'T3', 'start': 4, 'end': 10, 'duration': 6, 'demands': [2, 1]}, {'task': 'T4', 'start': 10, 'end': 14, 'duration': 4, 'demands': [3, 0]}, {'task': 'T5', 'start': 4, 'end': 10, 'duration': 6, 'demands': [1, 0]}, {'task': 'T6', 'start': 14, 'end': 19, 'duration': 5, 'demands': [2, 0]}, {'task': 'T7', 'start': 14, 'end': 18, 'duration': 4, 'demands': [1, 3]}, {'task': 'T8', 'start': 19, 'end': 20, 'duration': 1, 'demands': [0, 0]}]",30,csv,1
RCPSP,RCPSP,"Recently the ops crew sketched a release pipeline: a bunch of ordered tasks, each with a fixed duration and a need for some combination of compute and people while active. Some tasks are blocked until predecessors finish. The task now is to set start times for all tasks so the release completes at the earliest possible moment — in other words, minimize the time when the final task finishes. Schedules that make that finishing moment sooner are preferable, provided every dependency is respected, no task is left out or run twice, and resource caps for machines and staff aren’t exceeded. The exact tasks and resource numbers are shown below.

There are 9 tasks, 3 resource types, and resource capacities .

| resource_id | capacity |
|---|---|
| A | 7 |
| B | 7 |
| C | 5 |

| task_identifier | task_duration | resource_demands | successor_task_ids |
|---|---|---|---|
| T1 | 1 | 0, 0, 0 | T2, T3 |
| T2 | 1 | 2, 3, 1 | T4, T5 |
| T3 | 1 | 0, 4, 1 | T6 |
| T4 | 3 | 4, 6, 2 | T7 |
| T5 | 2 | 2, 1, 5 | T6 |
| T6 | 3 | 1, 2, 0 | T7 |
| T7 | 1 | 0, 0, 0 | none |
| T8 | 3 | 0, 0, 0 | none |
| T9 | 1 | 0, 0, 0 | T3 |

Assign start times so the release completes as early as possible while respecting dependencies and resource caps for all 9 tasks.

Also, when you hand me the schedule, please use this simple JSON shape so it's easy to read and plug into whatever checks we run next:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just means ""solution"" is a list of task entries; each entry has the task identifier and the numeric start and end times for that task (same time units as in the instance). It's just a template — replace the placeholders with the actual values for every task.

Please use the exact task identifiers from the instance input — don't rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.

Also be sure every task from the instance appears exactly once in the solution (no duplicates, none missing).","{'nr_tasks': 9, 'nr_resources': 3, 'capacities': [7, 7, 5], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2]}, {'duration': 1, 'demands': [2, 3, 1], 'successors': [3, 4]}, {'duration': 1, 'demands': [0, 4, 1], 'successors': [5]}, {'duration': 3, 'demands': [4, 6, 2], 'successors': [6]}, {'duration': 2, 'demands': [2, 1, 5], 'successors': [5]}, {'duration': 3, 'demands': [1, 2, 0], 'successors': [6]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 3, 'demands': [0, 0, 0], 'successors': []}, {'duration': 1, 'demands': [0, 0, 0], 'successors': [2]}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 1, 'end': 2, 'duration': 1, 'demands': [2, 3, 1]}, {'task': 2, 'start': 1, 'end': 2, 'duration': 1, 'demands': [0, 4, 1]}, {'task': 3, 'start': 4, 'end': 7, 'duration': 3, 'demands': [4, 6, 2]}, {'task': 4, 'start': 2, 'end': 4, 'duration': 2, 'demands': [2, 1, 5]}, {'task': 5, 'start': 7, 'end': 10, 'duration': 3, 'demands': [1, 2, 0]}, {'task': 6, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 8, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}]",11.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 7}, {'resource_id': 'B', 'capacity': 7}, {'resource_id': 'C', 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3']}, {'task_id': 'T2', 'duration': 1, 'demands': [2, 3, 1], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 1, 'demands': [0, 4, 1], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 3, 'demands': [4, 6, 2], 'successors': ['T7']}, {'task_id': 'T5', 'duration': 2, 'demands': [2, 1, 5], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 3, 'demands': [1, 2, 0], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 3, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T9', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T3']}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 1, 'end': 2, 'duration': 1, 'demands': [2, 3, 1]}, {'task': 'T3', 'start': 1, 'end': 2, 'duration': 1, 'demands': [0, 4, 1]}, {'task': 'T4', 'start': 4, 'end': 7, 'duration': 3, 'demands': [4, 6, 2]}, {'task': 'T5', 'start': 2, 'end': 4, 'duration': 2, 'demands': [2, 1, 5]}, {'task': 'T6', 'start': 7, 'end': 10, 'duration': 3, 'demands': [1, 2, 0]}, {'task': 'T7', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 0, 0]}, {'task': 'T9', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}]",31,markdown_table,names
RCPSP,RCPSP,"We’ve got a list of prep tasks for service: each one takes a fixed amount of time, uses some of the ovens and staff while it’s happening, and a few tasks only make sense after other tasks finish. The choice is simply about when to kick each task off so service can start sooner — measured by the moment the final prep completes, which we want to make as early as possible. At the same time, no oven or team member can be booked by more than one task at once, and every prep must be scheduled once in the right order. The actual task times and resource limits are shown below.

# nr_prep_tasks=4
# nr_resource_types=3
# resources=A, B, C
resource_id,capacity
A,7
B,7
C,3
prep_id,prep_duration,resource_demands,successor_prep_ids
T1,2,""4, 2, 1"",""T2, T4, T3""
T2,5,""2, 2, 3"",""T3, T4""
T3,2,""2, 2, 0"",""T4""
T4,1,""0, 0, 0"",""none""

Also, when you send the schedule back, please use this simple JSON layout so it's easy to parse — nothing fancy, just that shape.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This little sketch shows the shape I expect: ""solution"" is a list of the scheduled preps; each entry names the task and the time it starts and finishes. Think of ""task"" as the task label from the instance, and ""start""/""end"" as the time (in the same units used in the instance) when that prep runs. It's just an example of the shape, not the actual schedule.

Please make sure to use the exact identifiers from the instance input — don't rename them or invent new labels. For example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""

Replace the placeholders with those real ids and numeric start/end times when you submit the actual schedule.","{'nr_tasks': 4, 'nr_resources': 3, 'capacities': [7, 7, 3], 'tasks': [{'duration': 2, 'demands': [4, 2, 1], 'successors': [1, 3, 2]}, {'duration': 5, 'demands': [2, 2, 3], 'successors': [2, 3]}, {'duration': 2, 'demands': [2, 2, 0], 'successors': [3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [4, 2, 1]}, {'task': 1, 'start': 2, 'end': 7, 'duration': 5, 'demands': [2, 2, 3]}, {'task': 2, 'start': 7, 'end': 9, 'duration': 2, 'demands': [2, 2, 0]}, {'task': 3, 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 7}, {'resource_id': 'B', 'capacity': 7}, {'resource_id': 'C', 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [4, 2, 1], 'successors': ['T2', 'T4', 'T3']}, {'task_id': 'T2', 'duration': 5, 'demands': [2, 2, 3], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [2, 2, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [4, 2, 1]}, {'task': 'T2', 'start': 2, 'end': 7, 'duration': 5, 'demands': [2, 2, 3]}, {'task': 'T3', 'start': 7, 'end': 9, 'duration': 2, 'demands': [2, 2, 0]}, {'task': 'T4', 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",32,csv,names
RCPSP,RCPSP,"I picture the ground crew racing the clock: fuel, catering, cleaning and checks all have set lengths, a handful of trucks and people to go around, and some jobs that can’t start until others are done. The job here is to pick a start time for every ground task so that nothing is missed or done twice, no piece of equipment or staff is overbooked, and the plane can push back as soon as possible. The plane’s departure time is simply the moment the last required task wraps up, so the goal is to make that finishing moment as early as it can be. The exact list of activities, who’s available and how long each takes will be shown below.

Below are the 10 ground activities, the 3 resource types, and their capacities .
I see activity T1: it lasts 1, demands 0, 0, 0 while running, and must finish before its successors T2, T3, T5 can start.
I see activity T2: it lasts 2, demands 4, 4, 2 while running, and must finish before its successors T4, T5 can start.
I see activity T3: it lasts 4, demands 0, 1, 2 while running, and must finish before its successors T6 can start.
I see activity T4: it lasts 4, demands 5, 3, 5 while running, and must finish before its successors T7 can start.
I see activity T5: it lasts 4, demands 3, 1, 3 while running, and must finish before its successors T6, T9 can start.
I see activity T6: it lasts 2, demands 1, 2, 0 while running, and must finish before its successors T7, T8 can start.
I see activity T7: it lasts 1, demands 0, 0, 0 while running, and must finish before its successors none can start.
I see activity T8: it lasts 4, demands 1, 1, 0 while running, and must finish before its successors none can start.
I see activity T9: it lasts 1, demands 0, 0, 0 while running, and must finish before its successors T3 can start.
I see activity T10: it lasts 3, demands 1, 2, 0 while running, and must finish before its successors none can start.
I’ll use these details to assign start times so no equipment or staff is overbooked and the plane can push back as soon as possible.

When you send the schedule back, just use this simple JSON layout so it's easy to parse and check.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just means: ""solution"" is a list of scheduled activities; each item names the task, when it starts and when it finishes. Think of it like filling out a little form for every ground task — task is the task's id, start and end are the times (same time units as the instance). This snippet is only a sketch of the shape I need, not the actual completed schedule.

Please use the exact task identifiers from the instance input — do not rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”
- single capital letters like “A” or “B”
- or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 10, 'nr_resources': 3, 'capacities': [6, 7, 5], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2, 4]}, {'duration': 2, 'demands': [4, 4, 2], 'successors': [3, 4]}, {'duration': 4, 'demands': [0, 1, 2], 'successors': [5]}, {'duration': 4, 'demands': [5, 3, 5], 'successors': [6]}, {'duration': 4, 'demands': [3, 1, 3], 'successors': [5, 8]}, {'duration': 2, 'demands': [1, 2, 0], 'successors': [6, 7]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 4, 'demands': [1, 1, 0], 'successors': []}, {'duration': 1, 'demands': [0, 0, 0], 'successors': [2]}, {'duration': 3, 'demands': [1, 2, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [4, 4, 2]}, {'task': 2, 'start': 8, 'end': 12, 'duration': 4, 'demands': [0, 1, 2]}, {'task': 3, 'start': 12, 'end': 16, 'duration': 4, 'demands': [5, 3, 5]}, {'task': 4, 'start': 3, 'end': 7, 'duration': 4, 'demands': [3, 1, 3]}, {'task': 5, 'start': 12, 'end': 14, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 6, 'start': 17, 'end': 18, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 14, 'end': 18, 'duration': 4, 'demands': [1, 1, 0]}, {'task': 8, 'start': 7, 'end': 8, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 9, 'start': 0, 'end': 3, 'duration': 3, 'demands': [1, 2, 0]}]",18.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 6}, {'resource_id': 1, 'capacity': 7}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3', 'T5']}, {'task_id': 'T2', 'duration': 2, 'demands': [4, 4, 2], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 4, 'demands': [0, 1, 2], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 4, 'demands': [5, 3, 5], 'successors': ['T7']}, {'task_id': 'T5', 'duration': 4, 'demands': [3, 1, 3], 'successors': ['T6', 'T9']}, {'task_id': 'T6', 'duration': 2, 'demands': [1, 2, 0], 'successors': ['T7', 'T8']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 4, 'demands': [1, 1, 0], 'successors': []}, {'task_id': 'T9', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T3']}, {'task_id': 'T10', 'duration': 3, 'demands': [1, 2, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [4, 4, 2]}, {'task': 'T3', 'start': 8, 'end': 12, 'duration': 4, 'demands': [0, 1, 2]}, {'task': 'T4', 'start': 12, 'end': 16, 'duration': 4, 'demands': [5, 3, 5]}, {'task': 'T5', 'start': 3, 'end': 7, 'duration': 4, 'demands': [3, 1, 3]}, {'task': 'T6', 'start': 12, 'end': 14, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 'T7', 'start': 17, 'end': 18, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 14, 'end': 18, 'duration': 4, 'demands': [1, 1, 0]}, {'task': 'T9', 'start': 7, 'end': 8, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T10', 'start': 0, 'end': 3, 'duration': 3, 'demands': [1, 2, 0]}]",33,nl,0
RCPSP,RCPSP,"There’s a busy day at the printing house: a chain of production steps that all must happen, each taking a fixed stretch of time and grabbing some of the editors or a press while it’s active. The job is to set start times for each step so the entire run ends as soon as possible — the final completion moment is just the time the last step wraps up (check each step’s start plus its duration and take the latest). Certain steps can’t be started until others finish, and at no point can the team or presses in use exceed the shop’s available crew and machines. Every step must be carried out once, no skipping or repeating, and the detailed list of tasks and resource limits appears below.

It lists 4 production steps across 2 resource types, with shop capacities given as .
Step T1 runs for 1 time units, occupies 0, 3 resources while active, and must finish before T3, T4 may start.
Step T2 runs for 2 time units, occupies 2, 0 resources while active, and must finish before T3, T4, T1 may start.
Step T3 runs for 7 time units, occupies 5, 0 resources while active, and must finish before none may start.
Step T4 runs for 3 time units, occupies 1, 1 resources while active, and must finish before T3 may start.
All steps must be scheduled once so the final completion time is minimized while respecting the listed capacities and finish-to-start precedences.

When you send back the schedule, just use a tiny JSON snippet so it's easy to parse—nothing fancy, just the shape below.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that like a short form: ""solution"" holds a list of the steps, and each step says which task it is and when it starts and finishes. Keep it simple — task is the task identifier, start is the start time, and end is the end time. This block is just a sketch of the expected shape, not the actual schedule.

Please use the exact identifiers from the instance input — do not rename them or invent new ones. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 4, 'nr_resources': 2, 'capacities': [5, 4], 'tasks': [{'duration': 1, 'demands': [0, 3], 'successors': [2, 3]}, {'duration': 2, 'demands': [2, 0], 'successors': [2, 3, 0]}, {'duration': 7, 'demands': [5, 0], 'successors': []}, {'duration': 3, 'demands': [1, 1], 'successors': [2]}]}","[{'task': 0, 'start': 2, 'end': 3, 'duration': 1, 'demands': [0, 3]}, {'task': 1, 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 0]}, {'task': 2, 'start': 6, 'end': 13, 'duration': 7, 'demands': [5, 0]}, {'task': 3, 'start': 3, 'end': 6, 'duration': 3, 'demands': [1, 1]}]",13.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 3], 'successors': ['T3', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [2, 0], 'successors': ['T3', 'T4', 'T1']}, {'task_id': 'T3', 'duration': 7, 'demands': [5, 0], 'successors': []}, {'task_id': 'T4', 'duration': 3, 'demands': [1, 1], 'successors': ['T3']}]}","[{'task': 'T1', 'start': 2, 'end': 3, 'duration': 1, 'demands': [0, 3]}, {'task': 'T2', 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 0]}, {'task': 'T3', 'start': 6, 'end': 13, 'duration': 7, 'demands': [5, 0]}, {'task': 'T4', 'start': 3, 'end': 6, 'duration': 3, 'demands': [1, 1]}]",34,nl,1
RCPSP,RCPSP,"We have a little production to plan: several shoots and follow-up edits that must be slotted into a tight timetable. The decision to make is when each shoot and edit should start so the campaign finishes at the earliest possible moment — that finishing moment is just the end time of the last activity, so the goal is to make that end time as small as possible. Each item has a fixed runtime, eats up studio time or an editor while it’s happening, and some jobs can only begin after others finish. All activities must be included exactly once and resources can’t be overbooked. The exact durations, resource availability and task orderings are listed below.

# total_activities=7
# resource_types=1
# resources=1
resource_id,capacity
1,6
activity_id,duration_time_units,resource_demand_vector,immediate_successor_ids
T1,1,""0"",""T2, T6""
T2,5,""6"",""T4, T7""
T3,2,""4"",""T7, T1, T2, T6""
T4,4,""5"",""T5""
T5,6,""1"",""T7, T6""
T6,6,""2"",""T7""
T7,1,""0"",""none""

Also, when you send the schedule back, please use this simple JSON shape so everything lines up with the rest of the workflow:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little form: ""solution"" holds a list of all the activities, and each activity entry just names the task and gives its start and end times. Keep the task identifier exactly as it appears in the instance, and use the same time units the instance uses (so the numbers line up with the durations you were given). This JSON is just a sketch of the shape I need, not the actual scheduled times.

Please make sure all task identifiers are used exactly as they appear in the instance input — do not rename them or invent new labels.
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 7, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 5]}, {'duration': 5, 'demands': [6], 'successors': [3, 6]}, {'duration': 2, 'demands': [4], 'successors': [6, 0, 1, 5]}, {'duration': 4, 'demands': [5], 'successors': [4]}, {'duration': 6, 'demands': [1], 'successors': [6, 5]}, {'duration': 6, 'demands': [2], 'successors': [6]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 2, 'end': 3, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 3, 'end': 8, 'duration': 5, 'demands': [6]}, {'task': 2, 'start': 0, 'end': 2, 'duration': 2, 'demands': [4]}, {'task': 3, 'start': 8, 'end': 12, 'duration': 4, 'demands': [5]}, {'task': 4, 'start': 12, 'end': 18, 'duration': 6, 'demands': [1]}, {'task': 5, 'start': 18, 'end': 24, 'duration': 6, 'demands': [2]}, {'task': 6, 'start': 24, 'end': 25, 'duration': 1, 'demands': [0]}]",25.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T6']}, {'task_id': 'T2', 'duration': 5, 'demands': [6], 'successors': ['T4', 'T7']}, {'task_id': 'T3', 'duration': 2, 'demands': [4], 'successors': ['T7', 'T1', 'T2', 'T6']}, {'task_id': 'T4', 'duration': 4, 'demands': [5], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 6, 'demands': [1], 'successors': ['T7', 'T6']}, {'task_id': 'T6', 'duration': 6, 'demands': [2], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 2, 'end': 3, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 3, 'end': 8, 'duration': 5, 'demands': [6]}, {'task': 'T3', 'start': 0, 'end': 2, 'duration': 2, 'demands': [4]}, {'task': 'T4', 'start': 8, 'end': 12, 'duration': 4, 'demands': [5]}, {'task': 'T5', 'start': 12, 'end': 18, 'duration': 6, 'demands': [1]}, {'task': 'T6', 'start': 18, 'end': 24, 'duration': 6, 'demands': [2]}, {'task': 'T7', 'start': 24, 'end': 25, 'duration': 1, 'demands': [0]}]",35,csv,1
RCPSP,RCPSP,"In our facilities office the morning briefing sounded like a timing game: each room setup takes a set amount of time, ties up a certain number of custodians and pieces of equipment while it’s underway, and some setups can’t start until others are wrapped. The planner’s choice is which minute each setup begins so that, given the limited staff and gear, every room ends up ready as early as possible. You judge a plan by the finish times — pick the latest finish among all setups and that’s when the entire job is done; sooner is better. Must-haves are straightforward: schedule every setup exactly once, never use more resources at the same time than exist, and follow any “after this” ordering rules. The specific list of setups, their durations, and the resource limits follow below.

{
  ""num_setups"": 7,
  ""num_resource_types"": 2,
  ""resources"": [
    1,
    2
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 1,
      ""capacity"": 5
    },
    {
      ""resource_id"": 2,
      ""capacity"": 4
    }
  ],
  ""tasks"": [
    {
      ""setup_id"": ""T1"",
      ""setup_duration_minutes"": 3,
      ""setup_resource_requirements"": [
        0,
        2
      ],
      ""must_follow_setups"": [
        ""T2"",
        ""T4"",
        ""T5""
      ]
    },
    {
      ""setup_id"": ""T2"",
      ""setup_duration_minutes"": 2,
      ""setup_resource_requirements"": [
        1,
        3
      ],
      ""must_follow_setups"": [
        ""T3"",
        ""T5""
      ]
    },
    {
      ""setup_id"": ""T3"",
      ""setup_duration_minutes"": 3,
      ""setup_resource_requirements"": [
        2,
        0
      ],
      ""must_follow_setups"": [
        ""T5"",
        ""T6"",
        ""T7""
      ]
    },
    {
      ""setup_id"": ""T4"",
      ""setup_duration_minutes"": 12,
      ""setup_resource_requirements"": [
        1,
        0
      ],
      ""must_follow_setups"": [
        ""T6"",
        ""T3""
      ]
    },
    {
      ""setup_id"": ""T5"",
      ""setup_duration_minutes"": 3,
      ""setup_resource_requirements"": [
        5,
        0
      ],
      ""must_follow_setups"": [
        ""T7"",
        ""T6""
      ]
    },
    {
      ""setup_id"": ""T6"",
      ""setup_duration_minutes"": 6,
      ""setup_resource_requirements"": [
        2,
        3
      ],
      ""must_follow_setups"": [
        ""T7""
      ]
    },
    {
      ""setup_id"": ""T7"",
      ""setup_duration_minutes"": 1,
      ""setup_resource_requirements"": [
        0,
        0
      ],
      ""must_follow_setups"": []
    }
  ]
}

When you send the schedule back, please use a small JSON sketch like this so it's easy to read and check:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just shows a list called ""solution"" where each entry names a setup (the task), and the minute it starts and the minute it finishes. Super simple — think of ""task"" as the setup identifier from the briefing, and ""start"" / ""end"" as whole-minute times.

This is only the shape I expect, not the actual schedule — fill in one object per setup from the instance.

Please use the exact identifiers from the instance input — do not rename them or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 7, 'nr_resources': 2, 'capacities': [5, 4], 'tasks': [{'duration': 3, 'demands': [0, 2], 'successors': [1, 3, 4]}, {'duration': 2, 'demands': [1, 3], 'successors': [2, 4]}, {'duration': 3, 'demands': [2, 0], 'successors': [4, 5, 6]}, {'duration': 12, 'demands': [1, 0], 'successors': [5, 2]}, {'duration': 3, 'demands': [5, 0], 'successors': [6, 5]}, {'duration': 6, 'demands': [2, 3], 'successors': [6]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 2]}, {'task': 1, 'start': 3, 'end': 5, 'duration': 2, 'demands': [1, 3]}, {'task': 2, 'start': 15, 'end': 18, 'duration': 3, 'demands': [2, 0]}, {'task': 3, 'start': 3, 'end': 15, 'duration': 12, 'demands': [1, 0]}, {'task': 4, 'start': 18, 'end': 21, 'duration': 3, 'demands': [5, 0]}, {'task': 5, 'start': 21, 'end': 27, 'duration': 6, 'demands': [2, 3]}, {'task': 6, 'start': 27, 'end': 28, 'duration': 1, 'demands': [0, 0]}]",28.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 3, 'demands': [0, 2], 'successors': ['T2', 'T4', 'T5']}, {'task_id': 'T2', 'duration': 2, 'demands': [1, 3], 'successors': ['T3', 'T5']}, {'task_id': 'T3', 'duration': 3, 'demands': [2, 0], 'successors': ['T5', 'T6', 'T7']}, {'task_id': 'T4', 'duration': 12, 'demands': [1, 0], 'successors': ['T6', 'T3']}, {'task_id': 'T5', 'duration': 3, 'demands': [5, 0], 'successors': ['T7', 'T6']}, {'task_id': 'T6', 'duration': 6, 'demands': [2, 3], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 2]}, {'task': 'T2', 'start': 3, 'end': 5, 'duration': 2, 'demands': [1, 3]}, {'task': 'T3', 'start': 15, 'end': 18, 'duration': 3, 'demands': [2, 0]}, {'task': 'T4', 'start': 3, 'end': 15, 'duration': 12, 'demands': [1, 0]}, {'task': 'T5', 'start': 18, 'end': 21, 'duration': 3, 'demands': [5, 0]}, {'task': 'T6', 'start': 21, 'end': 27, 'duration': 6, 'demands': [2, 3]}, {'task': 'T7', 'start': 27, 'end': 28, 'duration': 1, 'demands': [0, 0]}]",36,json,1
RCPSP,RCPSP,"Someone on site is juggling the schedule for excavation, foundation, framing, roofing, plumbing, and finishing and has to decide when each should start. Each work item takes a set amount of time and needs a fixed number of hands (and sometimes a crane) while it’s underway. Excavation must be completed before foundation begins, and framing waits on the foundation. Every task must happen exactly once for its full duration, and at no time can the total workers or cranes in use exceed the crew available. The measure of success is straightforward: the earlier the last job finishes, the better — find each job’s finish by adding its duration to its start time, and the project finish is the latest of those. The concrete details are shown below.

Someone on site faces 5 tasks, must track 3 resource types, and has resource capacities .

| resource_id | capacity |
|---|---|
| A | 5 |
| B | 5 |
| C | 4 |

| task_label | task_duration | resource_demands | successor_tasks |
|---|---|---|---|
| T1 | 1 | 2, 1, 2 | T2, T3, T5, T4 |
| T2 | 2 | 2, 4, 2 | T5, T4 |
| T3 | 4 | 1, 1, 3 | T4, T5, T2 |
| T4 | 3 | 1, 2, 0 | T5 |
| T5 | 1 | 0, 0, 0 | none |

Someone on site should schedule each task so the overall project finish is as early as possible.

If you'd like to give me the schedule in a tidy, machine-friendly way, just follow this JSON shape — nothing fancy, just a list of tasks with their start and end times:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This sketch means: put each work item in the ""solution"" array, use the task identifier exactly as it appears in the instance, and record the numeric start time and the numeric end time (end = start + duration). It's just the expected shape — not the real answer — so fill in the actual task IDs and times when you submit the schedule.

Please don't rename or invent task IDs — use the identifiers from the instance exactly as given. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 5, 'nr_resources': 3, 'capacities': [5, 5, 4], 'tasks': [{'duration': 1, 'demands': [2, 1, 2], 'successors': [1, 2, 4, 3]}, {'duration': 2, 'demands': [2, 4, 2], 'successors': [4, 3]}, {'duration': 4, 'demands': [1, 1, 3], 'successors': [3, 4, 1]}, {'duration': 3, 'demands': [1, 2, 0], 'successors': [4]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [2, 1, 2]}, {'task': 1, 'start': 5, 'end': 7, 'duration': 2, 'demands': [2, 4, 2]}, {'task': 2, 'start': 1, 'end': 5, 'duration': 4, 'demands': [1, 1, 3]}, {'task': 3, 'start': 7, 'end': 10, 'duration': 3, 'demands': [1, 2, 0]}, {'task': 4, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}]",11.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': ['A', 'B', 'C'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 5}, {'resource_id': 'B', 'capacity': 5}, {'resource_id': 'C', 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [2, 1, 2], 'successors': ['T2', 'T3', 'T5', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [2, 4, 2], 'successors': ['T5', 'T4']}, {'task_id': 'T3', 'duration': 4, 'demands': [1, 1, 3], 'successors': ['T4', 'T5', 'T2']}, {'task_id': 'T4', 'duration': 3, 'demands': [1, 2, 0], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [2, 1, 2]}, {'task': 'T2', 'start': 5, 'end': 7, 'duration': 2, 'demands': [2, 4, 2]}, {'task': 'T3', 'start': 1, 'end': 5, 'duration': 4, 'demands': [1, 1, 3]}, {'task': 'T4', 'start': 7, 'end': 10, 'duration': 3, 'demands': [1, 2, 0]}, {'task': 'T5', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0, 0, 0]}]",37,markdown_table,names
RCPSP,RCPSP,"Many people think baking is just about recipes, but here it’s about timing: set start times for mixing, oven runs, cooling periods, decorating sessions, and packing so everything runs without overrunning the available ovens and workspace. Every action takes a fixed time, consumes oven or counter spots (which are limited), follows a required order, and must be done once. The point is to have the batch ready sooner rather than later — the finish time is the moment the last job ends, and the best plan is the one that makes that moment as early as possible. Specific durations and resource counts come next below.

# total_steps=6
# num_resource_types=1
# resources=A
resource_id,capacity
A,4
step_id,step_duration,resource_demands,successor_steps
T1,1,""0"",""T2, T5, T4""
T2,2,""2"",""T4, T6""
T3,1,""3"",""T6""
T4,5,""3"",""T6, T3, T5""
T5,2,""2"",""T6, T3""
T6,1,""0"",""none""

Also, to keep things tidy when you send the plan back, please follow this JSON layout so I can read it easily:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a tiny sketch of the shape I need: each object in the ""solution"" array is one action in the baking process, with ""task"" holding the task identifier, ""start"" the time it begins, and ""end"" the time it finishes. Think of it like a little form: one line per task, with when it starts and when it ends.

One more thing — the JSON above is only the expected format, not the actual schedule. And please use the exact task identifiers from the instance input, with no renaming and no new labels. Valid identifiers look like plain numbers such as ""1"" or ""23"", single capital letters like ""A"" or ""B"", or a capital letter followed by digits like ""A1"" or ""X7"".","{'nr_tasks': 6, 'nr_resources': 1, 'capacities': [4], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 4, 3]}, {'duration': 2, 'demands': [2], 'successors': [3, 5]}, {'duration': 1, 'demands': [3], 'successors': [5]}, {'duration': 5, 'demands': [3], 'successors': [5, 2, 4]}, {'duration': 2, 'demands': [2], 'successors': [5, 2]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [2]}, {'task': 2, 'start': 10, 'end': 11, 'duration': 1, 'demands': [3]}, {'task': 3, 'start': 3, 'end': 8, 'duration': 5, 'demands': [3]}, {'task': 4, 'start': 8, 'end': 10, 'duration': 2, 'demands': [2]}, {'task': 5, 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",12.0,"{'problem_type': 'RCPSP', 'nr_tasks': 6, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6'], 'resources': ['A'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T5', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [2], 'successors': ['T4', 'T6']}, {'task_id': 'T3', 'duration': 1, 'demands': [3], 'successors': ['T6']}, {'task_id': 'T4', 'duration': 5, 'demands': [3], 'successors': ['T6', 'T3', 'T5']}, {'task_id': 'T5', 'duration': 2, 'demands': [2], 'successors': ['T6', 'T3']}, {'task_id': 'T6', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [2]}, {'task': 'T3', 'start': 10, 'end': 11, 'duration': 1, 'demands': [3]}, {'task': 'T4', 'start': 3, 'end': 8, 'duration': 5, 'demands': [3]}, {'task': 'T5', 'start': 8, 'end': 10, 'duration': 2, 'demands': [2]}, {'task': 'T6', 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",38,csv,names
RCPSP,RCPSP,"A small dev team needed a plan: schedule design, coding, code review, integration testing, and deployment so each job starts at the right moment — each one has a fixed duration and uses a given amount of developer or tester hours while active. The decision is about start times that honor the order (implementation must finish before review, testing follows integration) and avoid booking more hours than the team has available on any day. The clear measure of success is the deployment finish time — earlier completion means a better schedule. Every task must be carried out once in sequence, and team-hours can’t be overused. The full, concrete details are shown below.

# total_number_of_tasks=9
# number_of_resource_types=3
# resources=1, 2, 3
resource_id,capacity
1,8
2,6
3,8
task_identifier,duration_days,daily_resource_demand_hours,immediate_successor_task_ids
T1,6,""2, 2, 1"",""T2""
T2,3,""8, 2, 1"",""T3""
T3,6,""2, 0, 5"",""T7""
T4,7,""5, 1, 1"",""T5""
T5,8,""2, 3, 2"",""T9""
T6,5,""2, 2, 2"",""T7""
T7,2,""2, 1, 0"",""T8""
T8,6,""4, 3, 4"",""T9, T5""
T9,1,""0, 0, 0"",""none""

Also, when you send back the schedule, please use a simple JSON layout like this so it's easy to parse:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just a quick sketch of the shape I expect: ""solution"" is a list of task entries; each entry names the task (the exact task identifier from the instance), and gives the start and end times for that task. Think of it like filling out a little form for each job: which job, when it starts, and when it finishes. The block above is only the expected format — not the actual schedule.

Please make sure to use the exact identifiers from the instance input — don't rename tasks or invent new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 9, 'nr_resources': 3, 'capacities': [8, 6, 8], 'tasks': [{'duration': 6, 'demands': [2, 2, 1], 'successors': [1]}, {'duration': 3, 'demands': [8, 2, 1], 'successors': [2]}, {'duration': 6, 'demands': [2, 0, 5], 'successors': [6]}, {'duration': 7, 'demands': [5, 1, 1], 'successors': [4]}, {'duration': 8, 'demands': [2, 3, 2], 'successors': [8]}, {'duration': 5, 'demands': [2, 2, 2], 'successors': [6]}, {'duration': 2, 'demands': [2, 1, 0], 'successors': [7]}, {'duration': 6, 'demands': [4, 3, 4], 'successors': [8, 4]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 6, 'duration': 6, 'demands': [2, 2, 1]}, {'task': 1, 'start': 6, 'end': 9, 'duration': 3, 'demands': [8, 2, 1]}, {'task': 2, 'start': 9, 'end': 15, 'duration': 6, 'demands': [2, 0, 5]}, {'task': 3, 'start': 9, 'end': 16, 'duration': 7, 'demands': [5, 1, 1]}, {'task': 4, 'start': 23, 'end': 31, 'duration': 8, 'demands': [2, 3, 2]}, {'task': 5, 'start': 0, 'end': 5, 'duration': 5, 'demands': [2, 2, 2]}, {'task': 6, 'start': 15, 'end': 17, 'duration': 2, 'demands': [2, 1, 0]}, {'task': 7, 'start': 17, 'end': 23, 'duration': 6, 'demands': [4, 3, 4]}, {'task': 8, 'start': 31, 'end': 32, 'duration': 1, 'demands': [0, 0, 0]}]",32.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': [1, 2, 3], 'resource_capacities': [{'resource_id': 1, 'capacity': 8}, {'resource_id': 2, 'capacity': 6}, {'resource_id': 3, 'capacity': 8}], 'task_records': [{'task_id': 'T1', 'duration': 6, 'demands': [2, 2, 1], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 3, 'demands': [8, 2, 1], 'successors': ['T3']}, {'task_id': 'T3', 'duration': 6, 'demands': [2, 0, 5], 'successors': ['T7']}, {'task_id': 'T4', 'duration': 7, 'demands': [5, 1, 1], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 8, 'demands': [2, 3, 2], 'successors': ['T9']}, {'task_id': 'T6', 'duration': 5, 'demands': [2, 2, 2], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 2, 'demands': [2, 1, 0], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 6, 'demands': [4, 3, 4], 'successors': ['T9', 'T5']}, {'task_id': 'T9', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 6, 'duration': 6, 'demands': [2, 2, 1]}, {'task': 'T2', 'start': 6, 'end': 9, 'duration': 3, 'demands': [8, 2, 1]}, {'task': 'T3', 'start': 9, 'end': 15, 'duration': 6, 'demands': [2, 0, 5]}, {'task': 'T4', 'start': 9, 'end': 16, 'duration': 7, 'demands': [5, 1, 1]}, {'task': 'T5', 'start': 23, 'end': 31, 'duration': 8, 'demands': [2, 3, 2]}, {'task': 'T6', 'start': 0, 'end': 5, 'duration': 5, 'demands': [2, 2, 2]}, {'task': 'T7', 'start': 15, 'end': 17, 'duration': 2, 'demands': [2, 1, 0]}, {'task': 'T8', 'start': 17, 'end': 23, 'duration': 6, 'demands': [4, 3, 4]}, {'task': 'T9', 'start': 31, 'end': 32, 'duration': 1, 'demands': [0, 0, 0]}]",39,csv,1
RCPSP,RCPSP,"Many people forget how many little moving parts a report needs: a diagnostic scan, image processing, a specialist’s review, and finally signing. The decision is to slot start times for those four steps so each waits for the previous to finish, given each step’s fixed length and the limited scanner and specialist hours that each consumes. A smarter plan is simply the one that yields an earlier final sign-off — measured by when that last step completes — and every step has to happen exactly once in order without ever booking more scanner or specialist time than exists. The instance details are right below.

It lists 8 sequential stages, 1 resource types, and resource capacities .
Stage T1 runs for 1 time units, consumes 0 of each resource while active, and must finish before T2, T7, T3 may start.
Stage T2 runs for 4 time units, consumes 5 of each resource while active, and must finish before T3, T5 may start.
Stage T3 runs for 1 time units, consumes 4 of each resource while active, and must finish before T4 may start.
Stage T4 runs for 3 time units, consumes 3 of each resource while active, and must finish before T8 may start.
Stage T5 runs for 1 time units, consumes 8 of each resource while active, and must finish before T6 may start.
Stage T6 runs for 5 time units, consumes 3 of each resource while active, and must finish before T8 may start.
Stage T7 runs for 3 time units, consumes 2 of each resource while active, and must finish before T8 may start.
Stage T8 runs for 1 time units, consumes 0 of each resource while active, and must finish before none may start.
Respect those 8 stages and the 1 capacities  so the solver can minimize the final sign-off time.

Oh, and when you send the schedule back, please use this simple JSON shape so I can read it easily:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This is just the form I expect: ""solution"" holds a list of the steps, and each step object has a task identifier plus its start and end times (use the same time units as in the instance). Think of it like a little form you’re filling out — one line per step showing when it begins and when it finishes.

This JSON is only a sketch of the expected shape, not the actual answer. The real schedule should list every task from the instance with concrete start and end numbers.

Also, a quick reminder: all identifiers must be used exactly as they appear in the instance input — no renaming and no new labels.

Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 8, 'nr_resources': 1, 'capacities': [8], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 6, 2]}, {'duration': 4, 'demands': [5], 'successors': [2, 4]}, {'duration': 1, 'demands': [4], 'successors': [3]}, {'duration': 3, 'demands': [3], 'successors': [7]}, {'duration': 1, 'demands': [8], 'successors': [5]}, {'duration': 5, 'demands': [3], 'successors': [7]}, {'duration': 3, 'demands': [2], 'successors': [7]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 5, 'duration': 4, 'demands': [5]}, {'task': 2, 'start': 6, 'end': 7, 'duration': 1, 'demands': [4]}, {'task': 3, 'start': 7, 'end': 10, 'duration': 3, 'demands': [3]}, {'task': 4, 'start': 5, 'end': 6, 'duration': 1, 'demands': [8]}, {'task': 5, 'start': 6, 'end': 11, 'duration': 5, 'demands': [3]}, {'task': 6, 'start': 1, 'end': 4, 'duration': 3, 'demands': [2]}, {'task': 7, 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",12.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [0], 'resource_capacities': [{'resource_id': 0, 'capacity': 8}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T7', 'T3']}, {'task_id': 'T2', 'duration': 4, 'demands': [5], 'successors': ['T3', 'T5']}, {'task_id': 'T3', 'duration': 1, 'demands': [4], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 3, 'demands': [3], 'successors': ['T8']}, {'task_id': 'T5', 'duration': 1, 'demands': [8], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 5, 'demands': [3], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 3, 'demands': [2], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 5, 'duration': 4, 'demands': [5]}, {'task': 'T3', 'start': 6, 'end': 7, 'duration': 1, 'demands': [4]}, {'task': 'T4', 'start': 7, 'end': 10, 'duration': 3, 'demands': [3]}, {'task': 'T5', 'start': 5, 'end': 6, 'duration': 1, 'demands': [8]}, {'task': 'T6', 'start': 6, 'end': 11, 'duration': 5, 'demands': [3]}, {'task': 'T7', 'start': 1, 'end': 4, 'duration': 3, 'demands': [2]}, {'task': 'T8', 'start': 11, 'end': 12, 'duration': 1, 'demands': [0]}]",40,nl,0
RCPSP,RCPSP,"I like to imagine a busy shop where five things need to happen in order — chassis assembly, then engine installation, then wiring, then testing, and finally certification. The choice to make is when to start each of those steps so they all run in a smooth sequence: some steps can’t start until specific earlier ones finish. Each step always takes a fixed amount of time and needs a certain number of lifts and technicians while it runs, and the lifts and techs are limited, so the schedule can’t ask for more people or lifts at once than exist. The better the plan, the sooner the vehicle is truly finished — measure that by the time the certification step finishes — and every step must be scheduled exactly once, no skipping or doubling up. The concrete details are shown below.

I list the 8 steps to schedule, the 2 resource types they share, and the resource capacities over time: .
For T1 I’d schedule it to run for 1 time units, during which it consumes 0, 0 of the resources, and it must finish before T2 can start.
For T2 I’d schedule it to run for 4 time units, during which it consumes 0, 5 of the resources, and it must finish before T3, T5, T6 can start.
For T3 I’d schedule it to run for 7 time units, during which it consumes 1, 2 of the resources, and it must finish before T4, T6 can start.
For T4 I’d schedule it to run for 2 time units, during which it consumes 4, 0 of the resources, and it must finish before T6, T7 can start.
For T5 I’d schedule it to run for 13 time units, during which it consumes 1, 0 of the resources, and it must finish before T7 can start.
For T6 I’d schedule it to run for 6 time units, during which it consumes 2, 0 of the resources, and it must finish before T8 can start.
For T7 I’d schedule it to run for 2 time units, during which it consumes 1, 3 of the resources, and it must finish before T8 can start.
For T8 I’d schedule it to run for 1 time units, during which it consumes 0, 0 of the resources, and it must finish before none can start.
I’ll use these 8 details and the  capacities to plan the schedule so the certification step finishes as early as possible.

Also, when you send the schedule back, it helps to follow a tiny JSON layout so it's easy to read — something casual and predictable like this:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just shows the shape I expect: ""solution"" is a list of task entries, each entry records which task it is and the start and end times for that task. Think of it as a simple form: one line per step, with when it begins and when it wraps up. It's just a sketch of the expected shape, not the actual schedule.

Please make sure every task identifier you use matches exactly what's in the instance input — do not rename identifiers or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 8, 'nr_resources': 2, 'capacities': [6, 5], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': [1]}, {'duration': 4, 'demands': [0, 5], 'successors': [2, 4, 5]}, {'duration': 7, 'demands': [1, 2], 'successors': [3, 5]}, {'duration': 2, 'demands': [4, 0], 'successors': [5, 6]}, {'duration': 13, 'demands': [1, 0], 'successors': [6]}, {'duration': 6, 'demands': [2, 0], 'successors': [7]}, {'duration': 2, 'demands': [1, 3], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 5]}, {'task': 2, 'start': 5, 'end': 12, 'duration': 7, 'demands': [1, 2]}, {'task': 3, 'start': 12, 'end': 14, 'duration': 2, 'demands': [4, 0]}, {'task': 4, 'start': 5, 'end': 18, 'duration': 13, 'demands': [1, 0]}, {'task': 5, 'start': 14, 'end': 20, 'duration': 6, 'demands': [2, 0]}, {'task': 6, 'start': 18, 'end': 20, 'duration': 2, 'demands': [1, 3]}, {'task': 7, 'start': 20, 'end': 21, 'duration': 1, 'demands': [0, 0]}]",21.0,"{'problem_type': 'RCPSP', 'nr_tasks': 8, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': ['T2']}, {'task_id': 'T2', 'duration': 4, 'demands': [0, 5], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 7, 'demands': [1, 2], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 2, 'demands': [4, 0], 'successors': ['T6', 'T7']}, {'task_id': 'T5', 'duration': 13, 'demands': [1, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 6, 'demands': [2, 0], 'successors': ['T8']}, {'task_id': 'T7', 'duration': 2, 'demands': [1, 3], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 1, 'end': 5, 'duration': 4, 'demands': [0, 5]}, {'task': 'T3', 'start': 5, 'end': 12, 'duration': 7, 'demands': [1, 2]}, {'task': 'T4', 'start': 12, 'end': 14, 'duration': 2, 'demands': [4, 0]}, {'task': 'T5', 'start': 5, 'end': 18, 'duration': 13, 'demands': [1, 0]}, {'task': 'T6', 'start': 14, 'end': 20, 'duration': 6, 'demands': [2, 0]}, {'task': 'T7', 'start': 18, 'end': 20, 'duration': 2, 'demands': [1, 3]}, {'task': 'T8', 'start': 20, 'end': 21, 'duration': 1, 'demands': [0, 0]}]",41,nl,1
RCPSP,RCPSP,"At the gallery there’s a little choreography to plan: hang the pieces, tweak the lights, attach labels, let the curator do an inspection, then do the opening run-through. Each of those steps takes a known amount of time and consumes a certain number of installers and lighting rigs while it runs, and some steps can’t start until others finish. The planner’s job is to pick start times for each step so each one happens once, the available people and equipment aren’t overused at any moment, and the whole setup completes as early as possible — judged by the clock time when the final step finishes. The specifics — durations, available installers/rigs, and who depends on whom — are listed below.

{
  ""total_steps"": 5,
  ""resource_types_count"": 2,
  ""resources"": [
    1,
    2
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 1,
      ""capacity"": 4
    },
    {
      ""resource_id"": 2,
      ""capacity"": 5
    }
  ],
  ""tasks"": [
    {
      ""step_id"": ""T1"",
      ""step_duration"": 3,
      ""resource_demands"": [
        0,
        5
      ],
      ""following_steps"": [
        ""T2"",
        ""T4"",
        ""T5"",
        ""T3""
      ]
    },
    {
      ""step_id"": ""T2"",
      ""step_duration"": 4,
      ""resource_demands"": [
        2,
        2
      ],
      ""following_steps"": [
        ""T3"",
        ""T4"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T3"",
      ""step_duration"": 4,
      ""resource_demands"": [
        3,
        0
      ],
      ""following_steps"": [
        ""T4"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T4"",
      ""step_duration"": 3,
      ""resource_demands"": [
        3,
        0
      ],
      ""following_steps"": []
    },
    {
      ""step_id"": ""T5"",
      ""step_duration"": 7,
      ""resource_demands"": [
        1,
        4
      ],
      ""following_steps"": [
        ""T4""
      ]
    }
  ]
}

When you’re ready to hand me the schedule, a simple JSON layout like the one below is perfect — it’s just a tiny form that says which task starts and when it ends.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a little checklist: ""solution"" is the list of steps (each step is an object), ""task"" is the task identifier from the instance, and ""start"" and ""end"" are when that step begins and finishes (using the same time units as the instance). This block is only a sketch of the shape I’m expecting, not your actual answer.

Please make sure to use the exact identifiers from the instance input — no renaming and no new labels. Valid identifiers look like plain numbers such as ""1"" or ""23"", single capital letters like ""A"" or ""B"", or a capital letter followed by digits like ""A1"" or ""X7"".","{'nr_tasks': 5, 'nr_resources': 2, 'capacities': [4, 5], 'tasks': [{'duration': 3, 'demands': [0, 5], 'successors': [1, 3, 4, 2]}, {'duration': 4, 'demands': [2, 2], 'successors': [2, 3, 4]}, {'duration': 4, 'demands': [3, 0], 'successors': [3, 4]}, {'duration': 3, 'demands': [3, 0], 'successors': []}, {'duration': 7, 'demands': [1, 4], 'successors': [3]}]}","[{'task': 0, 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 5]}, {'task': 1, 'start': 3, 'end': 7, 'duration': 4, 'demands': [2, 2]}, {'task': 2, 'start': 7, 'end': 11, 'duration': 4, 'demands': [3, 0]}, {'task': 3, 'start': 18, 'end': 21, 'duration': 3, 'demands': [3, 0]}, {'task': 4, 'start': 11, 'end': 18, 'duration': 7, 'demands': [1, 4]}]",21.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 4}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 3, 'demands': [0, 5], 'successors': ['T2', 'T4', 'T5', 'T3']}, {'task_id': 'T2', 'duration': 4, 'demands': [2, 2], 'successors': ['T3', 'T4', 'T5']}, {'task_id': 'T3', 'duration': 4, 'demands': [3, 0], 'successors': ['T4', 'T5']}, {'task_id': 'T4', 'duration': 3, 'demands': [3, 0], 'successors': []}, {'task_id': 'T5', 'duration': 7, 'demands': [1, 4], 'successors': ['T4']}]}","[{'task': 'T1', 'start': 0, 'end': 3, 'duration': 3, 'demands': [0, 5]}, {'task': 'T2', 'start': 3, 'end': 7, 'duration': 4, 'demands': [2, 2]}, {'task': 'T3', 'start': 7, 'end': 11, 'duration': 4, 'demands': [3, 0]}, {'task': 'T4', 'start': 18, 'end': 21, 'duration': 3, 'demands': [3, 0]}, {'task': 'T5', 'start': 11, 'end': 18, 'duration': 7, 'demands': [1, 4]}]",42,json,1
RCPSP,RCPSP,"Many people on the team are involved in a pipeline: scanning, metadata entry, quality checks, and finally releasing the digitized items. Every step takes a fixed time and needs a few scanners or catalogers while it’s happening, and some steps can’t start until others are finished. The decision is about when each step should begin so resources never exceed their limits, no task is skipped or duplicated, and the project’s over as soon as possible — that “over” moment is simply when the final task finishes (look at all end times and pick the latest). The specific timings and capacities are listed below.

{
  ""total_steps"": 7,
  ""resource_types_count"": 2,
  ""resources"": [
    0,
    1
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 0,
      ""capacity"": 5
    },
    {
      ""resource_id"": 1,
      ""capacity"": 4
    }
  ],
  ""tasks"": [
    {
      ""step_id"": ""T1"",
      ""step_duration"": 1,
      ""resource_usage"": [
        0,
        4
      ],
      ""downstream_steps"": [
        ""T2"",
        ""T4"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T2"",
      ""step_duration"": 6,
      ""resource_usage"": [
        2,
        4
      ],
      ""downstream_steps"": [
        ""T3"",
        ""T5"",
        ""T6""
      ]
    },
    {
      ""step_id"": ""T3"",
      ""step_duration"": 4,
      ""resource_usage"": [
        1,
        0
      ],
      ""downstream_steps"": [
        ""T5"",
        ""T6""
      ]
    },
    {
      ""step_id"": ""T4"",
      ""step_duration"": 5,
      ""resource_usage"": [
        1,
        0
      ],
      ""downstream_steps"": [
        ""T6"",
        ""T3""
      ]
    },
    {
      ""step_id"": ""T5"",
      ""step_duration"": 6,
      ""resource_usage"": [
        5,
        0
      ],
      ""downstream_steps"": [
        ""T7""
      ]
    },
    {
      ""step_id"": ""T6"",
      ""step_duration"": 8,
      ""resource_usage"": [
        2,
        4
      ],
      ""downstream_steps"": [
        ""T7""
      ]
    },
    {
      ""step_id"": ""T7"",
      ""step_duration"": 1,
      ""resource_usage"": [
        0,
        0
      ],
      ""downstream_steps"": []
    }
  ]
}

Also, when you send the schedule back, it's easiest if you stick to a tiny JSON shape like this:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that as a simple form: ""solution"" holds a list of the steps, each entry names which task it is and the start and end times you picked. It's just an example of the shape I expect, not the actual schedule.

Please make sure to use the exact identifiers from the instance input — do not rename them or invent new labels. Valid identifiers look like plain numbers such as ""1"" or ""23"", single capital letters like ""A"" or ""B"", or a capital letter followed by digits like ""A1"" or ""X7"".","{'nr_tasks': 7, 'nr_resources': 2, 'capacities': [5, 4], 'tasks': [{'duration': 1, 'demands': [0, 4], 'successors': [1, 3, 4]}, {'duration': 6, 'demands': [2, 4], 'successors': [2, 4, 5]}, {'duration': 4, 'demands': [1, 0], 'successors': [4, 5]}, {'duration': 5, 'demands': [1, 0], 'successors': [5, 2]}, {'duration': 6, 'demands': [5, 0], 'successors': [6]}, {'duration': 8, 'demands': [2, 4], 'successors': [6]}, {'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 4]}, {'task': 1, 'start': 1, 'end': 7, 'duration': 6, 'demands': [2, 4]}, {'task': 2, 'start': 7, 'end': 11, 'duration': 4, 'demands': [1, 0]}, {'task': 3, 'start': 1, 'end': 6, 'duration': 5, 'demands': [1, 0]}, {'task': 4, 'start': 19, 'end': 25, 'duration': 6, 'demands': [5, 0]}, {'task': 5, 'start': 11, 'end': 19, 'duration': 8, 'demands': [2, 4]}, {'task': 6, 'start': 25, 'end': 26, 'duration': 1, 'demands': [0, 0]}]",26.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [0, 1], 'resource_capacities': [{'resource_id': 0, 'capacity': 5}, {'resource_id': 1, 'capacity': 4}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 4], 'successors': ['T2', 'T4', 'T5']}, {'task_id': 'T2', 'duration': 6, 'demands': [2, 4], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 4, 'demands': [1, 0], 'successors': ['T5', 'T6']}, {'task_id': 'T4', 'duration': 5, 'demands': [1, 0], 'successors': ['T6', 'T3']}, {'task_id': 'T5', 'duration': 6, 'demands': [5, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 8, 'demands': [2, 4], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 4]}, {'task': 'T2', 'start': 1, 'end': 7, 'duration': 6, 'demands': [2, 4]}, {'task': 'T3', 'start': 7, 'end': 11, 'duration': 4, 'demands': [1, 0]}, {'task': 'T4', 'start': 1, 'end': 6, 'duration': 5, 'demands': [1, 0]}, {'task': 'T5', 'start': 19, 'end': 25, 'duration': 6, 'demands': [5, 0]}, {'task': 'T6', 'start': 11, 'end': 19, 'duration': 8, 'demands': [2, 4]}, {'task': 'T7', 'start': 25, 'end': 26, 'duration': 1, 'demands': [0, 0]}]",43,json,0
RCPSP,RCPSP,"We’re looking at a typical maintenance cycle: first a safety walk-through, then ordering parts, followed by replacing worn components, calibrating the gear, and a final verification pass. Each activity lasts a set amount of time and needs inspectors and a limited number of specialized tools while it’s happening, and some activities have to wait for others to finish. The plan is to set the start time for each activity so every item is done once (no skipping or doubling up), we never overbook inspectors or tools at the same time, and the entire cycle completes as early as possible — the finish time being when the last task ends. The exact task times, resource limits, and required order are listed below.

# total_tasks=9
# resource_types_count=1
# resources=A
resource_id,capacity
A,6
activity_id,activity_duration,activity_resource_demands,successor_activity_ids
T1,1,""0"",""T2, T3, T8, T4""
T2,2,""1"",""T5, T8""
T3,5,""2"",""T4, T6, T7""
T4,1,""2"",""T5""
T5,2,""6"",""T9""
T6,3,""6"",""T7""
T7,3,""2"",""T9""
T8,3,""1"",""T9, T4, T7""
T9,1,""0"",""none""

You can just hand the schedule back in a tiny JSON snippet like this — one ""solution"" array with an entry per activity:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This shows the shape I expect: each object in the solution list is one activity, ""task"" is that activity's identifier, ""start"" is when it begins, and ""end"" is when it finishes. Think of it like a simple to-do list with start/end times — nothing fancy. This is just a sketch of the expected format, not the actual schedule.

Please make sure to use the exact identifiers from the instance input — do not rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 9, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 2, 7, 3]}, {'duration': 2, 'demands': [1], 'successors': [4, 7]}, {'duration': 5, 'demands': [2], 'successors': [3, 5, 6]}, {'duration': 1, 'demands': [2], 'successors': [4]}, {'duration': 2, 'demands': [6], 'successors': [8]}, {'duration': 3, 'demands': [6], 'successors': [6]}, {'duration': 3, 'demands': [2], 'successors': [8]}, {'duration': 3, 'demands': [1], 'successors': [8, 3, 6]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 3, 'duration': 2, 'demands': [1]}, {'task': 2, 'start': 1, 'end': 6, 'duration': 5, 'demands': [2]}, {'task': 3, 'start': 9, 'end': 10, 'duration': 1, 'demands': [2]}, {'task': 4, 'start': 12, 'end': 14, 'duration': 2, 'demands': [6]}, {'task': 5, 'start': 6, 'end': 9, 'duration': 3, 'demands': [6]}, {'task': 6, 'start': 9, 'end': 12, 'duration': 3, 'demands': [2]}, {'task': 7, 'start': 3, 'end': 6, 'duration': 3, 'demands': [1]}, {'task': 8, 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}]",15.0,"{'problem_type': 'RCPSP', 'nr_tasks': 9, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9'], 'resources': ['A'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T3', 'T8', 'T4']}, {'task_id': 'T2', 'duration': 2, 'demands': [1], 'successors': ['T5', 'T8']}, {'task_id': 'T3', 'duration': 5, 'demands': [2], 'successors': ['T4', 'T6', 'T7']}, {'task_id': 'T4', 'duration': 1, 'demands': [2], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 2, 'demands': [6], 'successors': ['T9']}, {'task_id': 'T6', 'duration': 3, 'demands': [6], 'successors': ['T7']}, {'task_id': 'T7', 'duration': 3, 'demands': [2], 'successors': ['T9']}, {'task_id': 'T8', 'duration': 3, 'demands': [1], 'successors': ['T9', 'T4', 'T7']}, {'task_id': 'T9', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 3, 'duration': 2, 'demands': [1]}, {'task': 'T3', 'start': 1, 'end': 6, 'duration': 5, 'demands': [2]}, {'task': 'T4', 'start': 9, 'end': 10, 'duration': 1, 'demands': [2]}, {'task': 'T5', 'start': 12, 'end': 14, 'duration': 2, 'demands': [6]}, {'task': 'T6', 'start': 6, 'end': 9, 'duration': 3, 'demands': [6]}, {'task': 'T7', 'start': 9, 'end': 12, 'duration': 3, 'demands': [2]}, {'task': 'T8', 'start': 3, 'end': 6, 'duration': 3, 'demands': [1]}, {'task': 'T9', 'start': 14, 'end': 15, 'duration': 1, 'demands': [0]}]",44,csv,names
RCPSP,RCPSP,"Recently a shop had to plan five steps — primer, main coat, drying, touch-up, sealing — where each step has a fixed length and requires certain painters and rack spots while it’s running. The planner’s job is to set start times so the natural order is kept (some steps can’t begin until others end) and at no time are there more painters or racks in use than the shop has. The best way is the one that gets the entire job fully wrapped up earliest — compare schedules by the time the final sealing finishes. No task can be skipped or repeated; the exact durations and resource availability are given below.

{
  ""total_steps"": 5,
  ""resource_types"": 2,
  ""resources"": [
    1,
    2
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": 1,
      ""capacity"": 4
    },
    {
      ""resource_id"": 2,
      ""capacity"": 5
    }
  ],
  ""tasks"": [
    {
      ""step_id"": ""T1"",
      ""step_duration"": 2,
      ""step_demands"": [
        0,
        5
      ],
      ""successor_steps"": [
        ""T2"",
        ""T4"",
        ""T3"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T2"",
      ""step_duration"": 4,
      ""step_demands"": [
        1,
        3
      ],
      ""successor_steps"": [
        ""T3"",
        ""T4"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T3"",
      ""step_duration"": 2,
      ""step_demands"": [
        3,
        0
      ],
      ""successor_steps"": [
        ""T4"",
        ""T5""
      ]
    },
    {
      ""step_id"": ""T4"",
      ""step_duration"": 7,
      ""step_demands"": [
        2,
        0
      ],
      ""successor_steps"": []
    },
    {
      ""step_id"": ""T5"",
      ""step_duration"": 6,
      ""step_demands"": [
        1,
        2
      ],
      ""successor_steps"": [
        ""T4""
      ]
    }
  ]
}

When you send the final schedule back, just use a simple JSON shape like this — easy to read and copy-paste.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Here ""solution"" is a list of the scheduled steps. Each item says which task (that task id goes in task), and the numeric start and end times for that task. Think of it like filling out a little form for each step: which step, when it starts, when it finishes. This block is just a sketch of the shape I need, not the actual schedule.

Please make sure to use the exact identifiers from the instance input — no renaming and no new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 5, 'nr_resources': 2, 'capacities': [4, 5], 'tasks': [{'duration': 2, 'demands': [0, 5], 'successors': [1, 3, 2, 4]}, {'duration': 4, 'demands': [1, 3], 'successors': [2, 3, 4]}, {'duration': 2, 'demands': [3, 0], 'successors': [3, 4]}, {'duration': 7, 'demands': [2, 0], 'successors': []}, {'duration': 6, 'demands': [1, 2], 'successors': [3]}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 1, 'start': 2, 'end': 6, 'duration': 4, 'demands': [1, 3]}, {'task': 2, 'start': 6, 'end': 8, 'duration': 2, 'demands': [3, 0]}, {'task': 3, 'start': 14, 'end': 21, 'duration': 7, 'demands': [2, 0]}, {'task': 4, 'start': 8, 'end': 14, 'duration': 6, 'demands': [1, 2]}]",21.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 4}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [0, 5], 'successors': ['T2', 'T4', 'T3', 'T5']}, {'task_id': 'T2', 'duration': 4, 'demands': [1, 3], 'successors': ['T3', 'T4', 'T5']}, {'task_id': 'T3', 'duration': 2, 'demands': [3, 0], 'successors': ['T4', 'T5']}, {'task_id': 'T4', 'duration': 7, 'demands': [2, 0], 'successors': []}, {'task_id': 'T5', 'duration': 6, 'demands': [1, 2], 'successors': ['T4']}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 'T2', 'start': 2, 'end': 6, 'duration': 4, 'demands': [1, 3]}, {'task': 'T3', 'start': 6, 'end': 8, 'duration': 2, 'demands': [3, 0]}, {'task': 'T4', 'start': 14, 'end': 21, 'duration': 7, 'demands': [2, 0]}, {'task': 'T5', 'start': 8, 'end': 14, 'duration': 6, 'demands': [1, 2]}]",45,json,1
RCPSP,RCPSP,"Recently the lab filled up with overlapping student experiments, and the instructor needs to slot start times for each activity. Every step takes a fixed amount of time, requires some of our limited benches or instruments while it runs, and a few steps are chained behind others. The plan has to make sure every activity happens once, no bench is double-booked, and dependent steps wait their turn, so that all projects finish as soon as possible. To compare plans, simply check the finishing time of the last scheduled activity — a smaller finishing time means a tighter, better schedule. The concrete schedule inputs (tasks, durations, resources, dependencies) are shown below.

{
  ""num_activities"": 10,
  ""num_resource_types"": 2,
  ""resources"": [
    ""A"",
    ""B""
  ],
  ""resource_capacities"": [
    {
      ""resource_id"": ""A"",
      ""capacity"": 3
    },
    {
      ""resource_id"": ""B"",
      ""capacity"": 5
    }
  ],
  ""tasks"": [
    {
      ""activity_id"": ""T1"",
      ""activity_duration"": 1,
      ""resource_demands"": [
        0,
        0
      ],
      ""successor_activities"": []
    },
    {
      ""activity_id"": ""T2"",
      ""activity_duration"": 2,
      ""resource_demands"": [
        0,
        5
      ],
      ""successor_activities"": [
        ""T3"",
        ""T5"",
        ""T6""
      ]
    },
    {
      ""activity_id"": ""T3"",
      ""activity_duration"": 5,
      ""resource_demands"": [
        2,
        2
      ],
      ""successor_activities"": [
        ""T4"",
        ""T6""
      ]
    },
    {
      ""activity_id"": ""T4"",
      ""activity_duration"": 2,
      ""resource_demands"": [
        2,
        0
      ],
      ""successor_activities"": [
        ""T6""
      ]
    },
    {
      ""activity_id"": ""T5"",
      ""activity_duration"": 6,
      ""resource_demands"": [
        2,
        0
      ],
      ""successor_activities"": [
        ""T7""
      ]
    },
    {
      ""activity_id"": ""T6"",
      ""activity_duration"": 8,
      ""resource_demands"": [
        3,
        0
      ],
      ""successor_activities"": []
    },
    {
      ""activity_id"": ""T7"",
      ""activity_duration"": 3,
      ""resource_demands"": [
        1,
        3
      ],
      ""successor_activities"": [
        ""T8""
      ]
    },
    {
      ""activity_id"": ""T8"",
      ""activity_duration"": 1,
      ""resource_demands"": [
        0,
        0
      ],
      ""successor_activities"": []
    },
    {
      ""activity_id"": ""T9"",
      ""activity_duration"": 2,
      ""resource_demands"": [
        1,
        2
      ],
      ""successor_activities"": []
    },
    {
      ""activity_id"": ""T10"",
      ""activity_duration"": 3,
      ""resource_demands"": [
        3,
        3
      ],
      ""successor_activities"": []
    }
  ]
}

Oh, and when you send the schedule back, please use this simple JSON shape so it's easy to parse and compare:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just shows a list of scheduled activities: each entry assigns a task its start time and end time. Think of ""task"" as the exact activity label from the plan, and ""start""/""end"" as the times (in the same units used in the instance) when that activity begins and finishes. The block above is just a sketch of the expected shape — don't treat it as the actual schedule.

Please make sure you use the exact task identifiers given in the instance input — do not rename them or invent new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 10, 'nr_resources': 2, 'capacities': [3, 5], 'tasks': [{'duration': 1, 'demands': [0, 0], 'successors': []}, {'duration': 2, 'demands': [0, 5], 'successors': [2, 4, 5]}, {'duration': 5, 'demands': [2, 2], 'successors': [3, 5]}, {'duration': 2, 'demands': [2, 0], 'successors': [5]}, {'duration': 6, 'demands': [2, 0], 'successors': [6]}, {'duration': 8, 'demands': [3, 0], 'successors': []}, {'duration': 3, 'demands': [1, 3], 'successors': [7]}, {'duration': 1, 'demands': [0, 0], 'successors': []}, {'duration': 2, 'demands': [1, 2], 'successors': []}, {'duration': 3, 'demands': [3, 3], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 1, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 2, 'start': 8, 'end': 13, 'duration': 5, 'demands': [2, 2]}, {'task': 3, 'start': 13, 'end': 15, 'duration': 2, 'demands': [2, 0]}, {'task': 4, 'start': 2, 'end': 8, 'duration': 6, 'demands': [2, 0]}, {'task': 5, 'start': 15, 'end': 23, 'duration': 8, 'demands': [3, 0]}, {'task': 6, 'start': 8, 'end': 11, 'duration': 3, 'demands': [1, 3]}, {'task': 7, 'start': 23, 'end': 24, 'duration': 1, 'demands': [0, 0]}, {'task': 8, 'start': 2, 'end': 4, 'duration': 2, 'demands': [1, 2]}, {'task': 9, 'start': 23, 'end': 26, 'duration': 3, 'demands': [3, 3]}]",26.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': ['A', 'B'], 'resource_capacities': [{'resource_id': 'A', 'capacity': 3}, {'resource_id': 'B', 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0], 'successors': []}, {'task_id': 'T2', 'duration': 2, 'demands': [0, 5], 'successors': ['T3', 'T5', 'T6']}, {'task_id': 'T3', 'duration': 5, 'demands': [2, 2], 'successors': ['T4', 'T6']}, {'task_id': 'T4', 'duration': 2, 'demands': [2, 0], 'successors': ['T6']}, {'task_id': 'T5', 'duration': 6, 'demands': [2, 0], 'successors': ['T7']}, {'task_id': 'T6', 'duration': 8, 'demands': [3, 0], 'successors': []}, {'task_id': 'T7', 'duration': 3, 'demands': [1, 3], 'successors': ['T8']}, {'task_id': 'T8', 'duration': 1, 'demands': [0, 0], 'successors': []}, {'task_id': 'T9', 'duration': 2, 'demands': [1, 2], 'successors': []}, {'task_id': 'T10', 'duration': 3, 'demands': [3, 3], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0]}, {'task': 'T2', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 'T3', 'start': 8, 'end': 13, 'duration': 5, 'demands': [2, 2]}, {'task': 'T4', 'start': 13, 'end': 15, 'duration': 2, 'demands': [2, 0]}, {'task': 'T5', 'start': 2, 'end': 8, 'duration': 6, 'demands': [2, 0]}, {'task': 'T6', 'start': 15, 'end': 23, 'duration': 8, 'demands': [3, 0]}, {'task': 'T7', 'start': 8, 'end': 11, 'duration': 3, 'demands': [1, 3]}, {'task': 'T8', 'start': 23, 'end': 24, 'duration': 1, 'demands': [0, 0]}, {'task': 'T9', 'start': 2, 'end': 4, 'duration': 2, 'demands': [1, 2]}, {'task': 'T10', 'start': 23, 'end': 26, 'duration': 3, 'demands': [3, 3]}]",46,json,names
RCPSP,RCPSP,"Someone needs to line up the whole test campaign by setting start times for each run. Each run has a fixed duration and consumes one or more of the lab’s limited chambers or rigs while it’s running, and some runs are dependent on earlier ones finishing first. The aim is simple: get the last test done as early as possible — the campaign’s length is the time when that final run ends, and shorter is better. All tests must be scheduled exactly once and the equipment must never be oversubscribed. The specific runs, durations, dependencies and resource limits are shown below.

Here are the 5 test runs, the 2 equipment types, and the equipment capacities per type: .
Run T1 runs for 2 time units, consumes 0, 5 of the equipment, and must finish before successor runs T2, T4, T5 can start.
Run T2 runs for 5 time units, consumes 1, 2 of the equipment, and must finish before successor runs T3, T4 can start.
Run T3 runs for 2 time units, consumes 4, 0 of the equipment, and must finish before successor runs T4, T5 can start.
Run T4 runs for 3 time units, consumes 5, 0 of the equipment, and must finish before successor runs none can start.
Run T5 runs for 6 time units, consumes 2, 3 of the equipment, and must finish before successor runs T4 can start.
Someone must schedule each run exactly once without oversubscribing equipment; the aim is to complete the campaign (finish the last run) as early as possible.

Quick note on how I’d like the schedule returned: please use this simple JSON layout for the reply so it’s easy to read and check.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of that JSON like a little form: ""solution"" is the list of all the scheduled runs; each entry names the run (""task""), when it begins (""start"") and when it finishes (""end""). Keep it light — this block is just the shape I expect, not the actual schedule.

Please use the exact identifiers from the instance input — don’t rename anything and don’t introduce new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 5, 'nr_resources': 2, 'capacities': [5, 5], 'tasks': [{'duration': 2, 'demands': [0, 5], 'successors': [1, 3, 4]}, {'duration': 5, 'demands': [1, 2], 'successors': [2, 3]}, {'duration': 2, 'demands': [4, 0], 'successors': [3, 4]}, {'duration': 3, 'demands': [5, 0], 'successors': []}, {'duration': 6, 'demands': [2, 3], 'successors': [3]}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 1, 'start': 2, 'end': 7, 'duration': 5, 'demands': [1, 2]}, {'task': 2, 'start': 7, 'end': 9, 'duration': 2, 'demands': [4, 0]}, {'task': 3, 'start': 15, 'end': 18, 'duration': 3, 'demands': [5, 0]}, {'task': 4, 'start': 9, 'end': 15, 'duration': 6, 'demands': [2, 3]}]",18.0,"{'problem_type': 'RCPSP', 'nr_tasks': 5, 'nr_resources': 2, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5'], 'resources': [1, 2], 'resource_capacities': [{'resource_id': 1, 'capacity': 5}, {'resource_id': 2, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [0, 5], 'successors': ['T2', 'T4', 'T5']}, {'task_id': 'T2', 'duration': 5, 'demands': [1, 2], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [4, 0], 'successors': ['T4', 'T5']}, {'task_id': 'T4', 'duration': 3, 'demands': [5, 0], 'successors': []}, {'task_id': 'T5', 'duration': 6, 'demands': [2, 3], 'successors': ['T4']}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [0, 5]}, {'task': 'T2', 'start': 2, 'end': 7, 'duration': 5, 'demands': [1, 2]}, {'task': 'T3', 'start': 7, 'end': 9, 'duration': 2, 'demands': [4, 0]}, {'task': 'T4', 'start': 15, 'end': 18, 'duration': 3, 'demands': [5, 0]}, {'task': 'T5', 'start': 9, 'end': 15, 'duration': 6, 'demands': [2, 3]}]",47,nl,1
RCPSP,RCPSP,"There’s a planner at the plant who needs to line up every workstation’s start time: each job has a set duration, uses certain stations or tools while in progress, and can’t start until earlier steps are done. The planner must place every job once on the timeline so no tool is used by too many jobs at the same time and the handoffs happen in the right order. A good plan is the one where the final job finishes as early as possible — simply check the finishing time of the last task to compare plans. The specific jobs, their durations, dependencies, and resource limits are listed below.

# num_workstation_tasks=10
# num_resource_types=3
# resources=0, 1, 2
resource_id,capacity
0,6
1,6
2,3
workstation_task_id,task_duration,resource_requirements,dependent_successor_tasks
T1,1,""0, 0, 0"",""T2, T3, T6""
T2,2,""4, 3, 1"",""T4, T5""
T3,3,""0, 3, 1"",""T6, T5""
T4,3,""3, 2, 2"",""T7""
T5,5,""3, 2, 2"",""T6""
T6,3,""2, 1, 0"",""T7, T4""
T7,1,""0, 0, 0"",""none""
T8,2,""5, 2, 1"",""T4""
T9,2,""2, 3, 2"",""none""
T10,4,""0, 4, 1"",""T9""

Also, when you send the schedule back, please use this simple JSON layout so it's easy to parse and double-check.

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

This just shows the shape I need: a ""solution"" is a list of task entries, and each entry is a little timetable row — which task it is, when it starts, and when it finishes. Think of it like filling out a simple form: task, start time, end time. It's only a sketch of the expected shape, not the actual schedule itself.

Please make sure every identifier you use matches exactly what's in the instance input — no renaming, no new labels. Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 10, 'nr_resources': 3, 'capacities': [6, 6, 3], 'tasks': [{'duration': 1, 'demands': [0, 0, 0], 'successors': [1, 2, 5]}, {'duration': 2, 'demands': [4, 3, 1], 'successors': [3, 4]}, {'duration': 3, 'demands': [0, 3, 1], 'successors': [5, 4]}, {'duration': 3, 'demands': [3, 2, 2], 'successors': [6]}, {'duration': 5, 'demands': [3, 2, 2], 'successors': [5]}, {'duration': 3, 'demands': [2, 1, 0], 'successors': [6, 3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'duration': 2, 'demands': [5, 2, 1], 'successors': [3]}, {'duration': 2, 'demands': [2, 3, 2], 'successors': []}, {'duration': 4, 'demands': [0, 4, 1], 'successors': [8]}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 1, 'start': 2, 'end': 4, 'duration': 2, 'demands': [4, 3, 1]}, {'task': 2, 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 3, 1]}, {'task': 3, 'start': 12, 'end': 15, 'duration': 3, 'demands': [3, 2, 2]}, {'task': 4, 'start': 4, 'end': 9, 'duration': 5, 'demands': [3, 2, 2]}, {'task': 5, 'start': 9, 'end': 12, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 6, 'start': 15, 'end': 16, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 7, 'start': 0, 'end': 2, 'duration': 2, 'demands': [5, 2, 1]}, {'task': 8, 'start': 9, 'end': 11, 'duration': 2, 'demands': [2, 3, 2]}, {'task': 9, 'start': 4, 'end': 8, 'duration': 4, 'demands': [0, 4, 1]}]",16.0,"{'problem_type': 'RCPSP', 'nr_tasks': 10, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7', 'T8', 'T9', 'T10'], 'resources': [0, 1, 2], 'resource_capacities': [{'resource_id': 0, 'capacity': 6}, {'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 3}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0, 0, 0], 'successors': ['T2', 'T3', 'T6']}, {'task_id': 'T2', 'duration': 2, 'demands': [4, 3, 1], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 3, 'demands': [0, 3, 1], 'successors': ['T6', 'T5']}, {'task_id': 'T4', 'duration': 3, 'demands': [3, 2, 2], 'successors': ['T7']}, {'task_id': 'T5', 'duration': 5, 'demands': [3, 2, 2], 'successors': ['T6']}, {'task_id': 'T6', 'duration': 3, 'demands': [2, 1, 0], 'successors': ['T7', 'T4']}, {'task_id': 'T7', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}, {'task_id': 'T8', 'duration': 2, 'demands': [5, 2, 1], 'successors': ['T4']}, {'task_id': 'T9', 'duration': 2, 'demands': [2, 3, 2], 'successors': []}, {'task_id': 'T10', 'duration': 4, 'demands': [0, 4, 1], 'successors': ['T9']}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T2', 'start': 2, 'end': 4, 'duration': 2, 'demands': [4, 3, 1]}, {'task': 'T3', 'start': 1, 'end': 4, 'duration': 3, 'demands': [0, 3, 1]}, {'task': 'T4', 'start': 12, 'end': 15, 'duration': 3, 'demands': [3, 2, 2]}, {'task': 'T5', 'start': 4, 'end': 9, 'duration': 5, 'demands': [3, 2, 2]}, {'task': 'T6', 'start': 9, 'end': 12, 'duration': 3, 'demands': [2, 1, 0]}, {'task': 'T7', 'start': 15, 'end': 16, 'duration': 1, 'demands': [0, 0, 0]}, {'task': 'T8', 'start': 0, 'end': 2, 'duration': 2, 'demands': [5, 2, 1]}, {'task': 'T9', 'start': 9, 'end': 11, 'duration': 2, 'demands': [2, 3, 2]}, {'task': 'T10', 'start': 4, 'end': 8, 'duration': 4, 'demands': [0, 4, 1]}]",48,csv,0
RCPSP,RCPSP,"I’m helping plan a celebration where a handful of setup crews and catering teams all have to get their jobs done in the right order. The planners need to pick a start time for every setup and food task so that things that depend on others don’t start too early, each job runs for its given length and uses the staff and gear assigned to it, and no team or piece of equipment is asked to do more than it has capacity for. The better plan is the one that gets the whole venue ready the fastest — in practice that means looking at when the last job finishes and making that time as early as possible (or, if you prefer, shrinking the time span from the first start to that final finish). Every task has to appear exactly once, can’t be split or duplicated, and the exact list of jobs, durations, staff and equipment available are shown below.

There are 7 jobs to schedule, using 1 resource types with capacities .

| resource_id | capacity |
|---|---|
| 1 | 6 |

| job_id | job_duration | resource_demands | successor_jobs |
|---|---|---|---|
| T1 | 1 | 0 | T2, T6, T7, T4 |
| T2 | 3 | 2 | T4, T5 |
| T3 | 1 | 4 | T7 |
| T4 | 1 | 6 | T5 |
| T5 | 3 | 4 | T7, T3 |
| T6 | 4 | 2 | T7, T5 |
| T7 | 1 | 0 | none |

Each job must appear exactly once; we’ll use these lines to build the fastest feasible plan.

You can just send the schedule back in a tiny JSON shape like this — nothing fancy, just a list of tasks with their start and end times:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of it like a little table: each object in the ""solution"" list is one job, ""task"" is the task identifier from the instance, and ""start"" and ""end"" are the times when that job begins and finishes. This is only a sketch of the expected shape — not the actual schedule — so don’t fill it in here unless you’re ready to submit the real timings.

Please use the exact task identifiers from the instance input — no renaming and no new labels. 
- for example: ""Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.""","{'nr_tasks': 7, 'nr_resources': 1, 'capacities': [6], 'tasks': [{'duration': 1, 'demands': [0], 'successors': [1, 5, 6, 3]}, {'duration': 3, 'demands': [2], 'successors': [3, 4]}, {'duration': 1, 'demands': [4], 'successors': [6]}, {'duration': 1, 'demands': [6], 'successors': [4]}, {'duration': 3, 'demands': [4], 'successors': [6, 2]}, {'duration': 4, 'demands': [2], 'successors': [6, 4]}, {'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 1, 'start': 1, 'end': 4, 'duration': 3, 'demands': [2]}, {'task': 2, 'start': 9, 'end': 10, 'duration': 1, 'demands': [4]}, {'task': 3, 'start': 5, 'end': 6, 'duration': 1, 'demands': [6]}, {'task': 4, 'start': 6, 'end': 9, 'duration': 3, 'demands': [4]}, {'task': 5, 'start': 1, 'end': 5, 'duration': 4, 'demands': [2]}, {'task': 6, 'start': 10, 'end': 11, 'duration': 1, 'demands': [0]}]",11.0,"{'problem_type': 'RCPSP', 'nr_tasks': 7, 'nr_resources': 1, 'tasks': ['T1', 'T2', 'T3', 'T4', 'T5', 'T6', 'T7'], 'resources': [1], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}], 'task_records': [{'task_id': 'T1', 'duration': 1, 'demands': [0], 'successors': ['T2', 'T6', 'T7', 'T4']}, {'task_id': 'T2', 'duration': 3, 'demands': [2], 'successors': ['T4', 'T5']}, {'task_id': 'T3', 'duration': 1, 'demands': [4], 'successors': ['T7']}, {'task_id': 'T4', 'duration': 1, 'demands': [6], 'successors': ['T5']}, {'task_id': 'T5', 'duration': 3, 'demands': [4], 'successors': ['T7', 'T3']}, {'task_id': 'T6', 'duration': 4, 'demands': [2], 'successors': ['T7', 'T5']}, {'task_id': 'T7', 'duration': 1, 'demands': [0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 1, 'duration': 1, 'demands': [0]}, {'task': 'T2', 'start': 1, 'end': 4, 'duration': 3, 'demands': [2]}, {'task': 'T3', 'start': 9, 'end': 10, 'duration': 1, 'demands': [4]}, {'task': 'T4', 'start': 5, 'end': 6, 'duration': 1, 'demands': [6]}, {'task': 'T5', 'start': 6, 'end': 9, 'duration': 3, 'demands': [4]}, {'task': 'T6', 'start': 1, 'end': 5, 'duration': 4, 'demands': [2]}, {'task': 'T7', 'start': 10, 'end': 11, 'duration': 1, 'demands': [0]}]",49,markdown_table,1
RCPSP,RCPSP,"Recently the engineering crew faced a staged rollout where every stage runs for a known stretch of time and consumes some servers and network capacity while active, and certain stages have to wait for earlier ones to finish. The planner’s job was to pick start moments for each stage so the whole rollout completes as quickly as possible — success is simply the finish time of the final stage, so a smaller finish time is better. It’s important that every stage appears once and only once, no stage begins before its prerequisites are done, and the combined server and bandwidth usage at any second never exceeds what’s available. The detailed instance (tasks, times, resources) follows below.

There are 4 stages, 3 resource types, and the resource capacities are .
Stage T1 runs for 2 time units, consumes 2, 2, 1 while active, and must finish before its successor stages T2, T3, T4 may start.
Stage T2 runs for 5 time units, consumes 2, 2, 5 while active, and must finish before its successor stages T3, T4 may start.
Stage T3 runs for 2 time units, consumes 1, 2, 0 while active, and must finish before its successor stages T4 may start.
Stage T4 runs for 1 time units, consumes 0, 0, 0 while active, and must finish before its successor stages none may start.
The planner must schedule every stage once so the rollout finishes as quickly as possible without exceeding the capacities .

Also, to keep things tidy when you send back the schedule, please follow this simple JSON layout — just a little template to show how I expect the answer to be shaped:

{
  ""solution"": [
    {
      ""task"": <task_id>,
      ""start"": <start_time>,
      ""end"": <end_time>
    }
  ]
}

Think of this like a little form: ""solution"" is a list of stages, and each entry says which task it is and when it begins and finishes. Super casual — just fill in each task once with its start and end times from the instance. This block is only a sketch of the shape I want, not the actual schedule.

Please be sure to use the exact identifiers as they appear in the instance input — do not rename tasks or invent new labels. 

Valid identifiers look like plain numbers such as “1” or “23”, single capital letters like “A” or “B”, or a capital letter followed by digits like “A1” or “X7”.","{'nr_tasks': 4, 'nr_resources': 3, 'capacities': [6, 4, 5], 'tasks': [{'duration': 2, 'demands': [2, 2, 1], 'successors': [1, 2, 3]}, {'duration': 5, 'demands': [2, 2, 5], 'successors': [2, 3]}, {'duration': 2, 'demands': [1, 2, 0], 'successors': [3]}, {'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 0, 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 2, 1]}, {'task': 1, 'start': 2, 'end': 7, 'duration': 5, 'demands': [2, 2, 5]}, {'task': 2, 'start': 7, 'end': 9, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 3, 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",10.0,"{'problem_type': 'RCPSP', 'nr_tasks': 4, 'nr_resources': 3, 'tasks': ['T1', 'T2', 'T3', 'T4'], 'resources': [1, 2, 3], 'resource_capacities': [{'resource_id': 1, 'capacity': 6}, {'resource_id': 2, 'capacity': 4}, {'resource_id': 3, 'capacity': 5}], 'task_records': [{'task_id': 'T1', 'duration': 2, 'demands': [2, 2, 1], 'successors': ['T2', 'T3', 'T4']}, {'task_id': 'T2', 'duration': 5, 'demands': [2, 2, 5], 'successors': ['T3', 'T4']}, {'task_id': 'T3', 'duration': 2, 'demands': [1, 2, 0], 'successors': ['T4']}, {'task_id': 'T4', 'duration': 1, 'demands': [0, 0, 0], 'successors': []}]}","[{'task': 'T1', 'start': 0, 'end': 2, 'duration': 2, 'demands': [2, 2, 1]}, {'task': 'T2', 'start': 2, 'end': 7, 'duration': 5, 'demands': [2, 2, 5]}, {'task': 'T3', 'start': 7, 'end': 9, 'duration': 2, 'demands': [1, 2, 0]}, {'task': 'T4', 'start': 9, 'end': 10, 'duration': 1, 'demands': [0, 0, 0]}]",50,nl,1