anonymous commited on
Commit
8952689
·
1 Parent(s): 3098dca
Files changed (1) hide show
  1. README.md +34 -0
README.md CHANGED
@@ -1,3 +1,37 @@
1
  ---
2
  license: cc-by-nc-4.0
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-4.0
3
  ---
4
+
5
+ # SURF: A Generalisation Benchmark for GNNs Predicting Fluid Dynamics
6
+
7
+ SURF, is a benchmark designed to test the generalization of learned graph-based fluid simulators. The benchmark consists of seven independent datasets:
8
+
9
+ - Base
10
+ - Turned
11
+ - Topo
12
+ - Range
13
+ - Dynamic
14
+ - Full
15
+ - FullFiner
16
+
17
+ Each dataset is available as separate *.zip file and consists of at least 1200 2D incompressible fluid flow simulations with 300 timesteps.
18
+ The data structure is as follows:
19
+ - folder: dataset_name
20
+ - folders: dpx
21
+ - files: sim.npz, triangles.py, constrained_kmeans_20.npy, Simulation_dp1_Timestep_50.png
22
+ - folder: Splits
23
+ - files: train.txt, test.txt, valid.txt
24
+
25
+ The file sim.npz (numpy archive) contains the result of the simulation for each timestep at each node:
26
+ - 'pointcloud': x, y coordinates
27
+ - 'VX': velocity in x-direction
28
+ - 'VY': velocity in y-direction
29
+ - 'PS': static pressure
30
+ - 'PG': dynamic pressure
31
+ - 'T': temperature
32
+ - 'TC': thermal conductivity of fluid
33
+ - 'HC': heat capacity of fluid
34
+
35
+ The results have the following shape: VX.shape=(#timesteps, #nodes, 1).
36
+
37
+ The file triangles.py contains the mesh connectivity. triangles.shape=(#timesteps, #elements, 3). Each triangle is defined by the node numbers in counter clockwise direction.