File size: 7,854 Bytes
3193174
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
"""
Basic framework usage β€” no LLM required.

Demonstrates the core building blocks of gMAS:
  1. Building a property graph from AgentProfile objects
  2. Dynamic topology changes (add / remove edges, update adjacency)
  3. Topological execution order and parallel groups
  4. Decentralised agent state management
  5. Encoding agent profiles into embeddings
  6. Converting to PyG Data for GNN training
  7. Extracting subgraphs

Run:
    python -m examples.basic_usage
"""

import torch

from builder import build_property_graph
from core.agent import AgentProfile
from core.encoder import NodeEncoder
from execution import build_execution_order, get_parallel_groups

# ── Helpers ──────────────────────────────────────────────────────────────────


def _header(title: str) -> None:
    print(f"\n{'─' * 50}\n  {title}\n{'─' * 50}")


# ── 1. Building a property graph ────────────────────────────────────────────


def example_basic_graph():
    """Create a three-agent graph with a shared task node."""
    agents = [
        AgentProfile(
            agent_id="math_solver",
            display_name="Math Solver",
            description="Solves mathematical problems step by step",
            tools=["calculator"],
        ),
        AgentProfile(
            agent_id="code_writer",
            display_name="Code Writer",
            description="Writes Python code to solve problems",
            tools=["python", "code_execution"],
        ),
        AgentProfile(
            agent_id="checker",
            display_name="Answer Checker",
            description="Verifies the correctness of solutions",
        ),
    ]

    edges = [("math_solver", "checker"), ("code_writer", "checker")]

    graph = build_property_graph(
        agents,
        workflow_edges=edges,
        query="What is 25 * 17?",
        answer="425",
        include_task_node=True,
    )

    print(f"  Agents : {[a.agent_id for a in agents]}")
    print(f"  Edges  : {edges}")
    print(f"  Query  : {graph.query}")
    return graph


# ── 2. Dynamic topology ─────────────────────────────────────────────────────


def example_dynamic_topology():
    """Add / remove edges and push a new adjacency matrix at runtime."""
    graph = example_basic_graph()

    graph.add_edge("math_solver", "code_writer", weight=0.8)
    print("  + edge math_solver β†’ code_writer (weight 0.8)")

    graph.remove_edge("math_solver", "code_writer")
    print("  βˆ’ edge math_solver β†’ code_writer removed")

    new_adj = torch.tensor(
        [[0, 1, 1, 0], [0, 0, 1, 0], [0, 0, 0, 0], [1, 1, 1, 0]],
        dtype=torch.float32,
    )
    graph.update_communication(new_adj)
    print(f"  Adjacency matrix updated β€” shape {new_adj.shape}")
    return graph


# ── 3. Execution order ──────────────────────────────────────────────────────


def example_execution_order():
    """Compute topological order and parallel groups for a diamond DAG."""
    agents = [
        AgentProfile(agent_id="a", display_name="Agent A"),
        AgentProfile(agent_id="b", display_name="Agent B"),
        AgentProfile(agent_id="c", display_name="Agent C"),
        AgentProfile(agent_id="d", display_name="Agent D"),
    ]
    edges = [("a", "b"), ("a", "c"), ("b", "d"), ("c", "d")]

    graph = build_property_graph(agents, workflow_edges=edges, include_task_node=False)
    ids = [a.agent_id for a in agents]

    order = build_execution_order(graph.A_com, ids)
    groups = get_parallel_groups(graph.A_com, ids)

    print(f"  Order  : {order}")
    print(f"  Groups : {groups}")
    print("  β†’ b and c can run in parallel before d")
    return graph


# ── 4. Decentralised state ──────────────────────────────────────────────────


# ── 4. Decentralised state ──────────────────────────────────────────────────


def example_decentralised_state():
    """Immutable append / clear on an AgentProfile's local state."""
    agent = AgentProfile(
        agent_id="assistant",
        display_name="Assistant",
        state=[{"role": "system", "content": "You are a helpful assistant."}],
    )

    agent = agent.append_state({"role": "user", "content": "Hello!"})
    agent = agent.append_state({"role": "assistant", "content": "Hi there!"})
    print(f"  After 2 turns : {len(agent.state)} messages")

    cleared = agent.clear_state()
    print(f"  After clear   : {len(cleared.state)} messages")
    return agent


# ── 5. Embeddings ───────────────────────────────────────────────────────────


def example_embeddings():
    """Encode agent descriptions into fixed-size hash embeddings."""
    encoder = NodeEncoder(model_name="hash:128")

    agents = [
        AgentProfile(
            agent_id="solver",
            display_name="Math Solver",
            description="Expert in mathematics and calculations",
        ),
        AgentProfile(
            agent_id="writer",
            display_name="Code Writer",
            description="Expert in Python programming",
        ),
    ]

    embeddings = encoder.encode([a.to_text() for a in agents])
    print(f"  Shape: {embeddings.shape}")  # (2, 128)

    agents_emb = [a.with_embedding(embeddings[i]) for i, a in enumerate(agents)]

    return build_property_graph(
        agents_emb,
        workflow_edges=[("solver", "writer")],
        include_task_node=False,
    )


# ── 6. PyG conversion ──────────────────────────────────────────────────────


def example_pyg_conversion():
    """Convert a property graph to a PyTorch Geometric Data object."""
    graph = example_embeddings()
    data = graph.to_pyg_data()
    print(f"  Node features : {data.x.shape}")
    print(f"  Edge index    : {data.edge_index.shape}")
    return data


# ── 7. Subgraph extraction ─────────────────────────────────────────────────


def example_subgraph():
    """Extract a subgraph containing only a subset of agents."""
    graph = example_basic_graph()
    sub = graph.subgraph(["math_solver", "checker"])
    print(f"  Original : {[a.agent_id for a in graph.agents]}")
    print(f"  Subgraph : {[a.agent_id for a in sub.agents]}")
    return sub


# ── Entry point ─────────────────────────────────────────────────────────────


def main():
    examples = [
        ("1. Basic graph", example_basic_graph),
        ("2. Dynamic topology", example_dynamic_topology),
        ("3. Execution order", example_execution_order),
        ("4. Decentralised state", example_decentralised_state),
        ("5. Embeddings", example_embeddings),
        ("6. PyG conversion", example_pyg_conversion),
        ("7. Subgraph extraction", example_subgraph),
    ]

    for title, fn in examples:
        _header(title)
        fn()

    print("\nAll examples completed βœ…")


if __name__ == "__main__":
    main()