def calculate_efficiency(gene_nodes_pos, gene_edges_thickness, gene_adj_element, np_save_path=False):
condition_nodes_pos, input_nodes, input_vectors, output_nodes, \
output_vectors, frozen_nodes, condition_edges_indices, condition_edges_thickness\
= make_main_node_edge_info(*condition(), condition_edge_thickness=0.2)
# make edge_indices
edges_indices = make_adj_triu_matrix(gene_adj_element, node_num, condition_edges_indices)
# make nodes_pos
nodes_pos = np.concatenate([condition_nodes_pos, gene_nodes_pos])
# 条件ノードが含まれている部分グラフを抽出
G = nx.Graph()
G.add_nodes_from(np.arange(len(nodes_pos)))
G.add_edges_from(edges_indices)
condition_node_list = input_nodes + output_nodes + frozen_nodes
trigger = 0 # 条件ノードが全て接続するグラフが存在するとき，トリガーを発動する
for c in nx.connected_components(G):
sg = G.subgraph(c) # 部分グラフ
if set(condition_node_list) <= set(sg.nodes): # 条件ノードが全て含まれているか
edges_indices = np.array(sg.edges)
trigger = 1
break
if trigger == 0: # ペナルティを発動する
return -10.0
# make edges_thickness
edges_thickness = make_edge_thick_triu_matrix(gene_edges_thickness, node_num, condition_edges_indices, condition_edges_thickness, edges_indices)
env = BarFemGym(nodes_pos, input_nodes, input_vectors,
output_nodes, output_vectors, frozen_nodes,
edges_indices, edges_thickness, frozen_nodes)
env.reset()
efficiency = env.calculate_simulation()
if np_save_path:
env.render(save_path=os.path.join(np_save_path, "image.png"))
np.save(os.path.join(np_save_path, "nodes_pos.npy"), nodes_pos)
np.save(os.path.join(np_save_path, "edges_indices.npy"), edges_indices)
np.save(os.path.join(np_save_path, "edges_thickness.npy"), edges_thickness)
return float(efficiency)
save_graph_info(os.path.join(log_dir, 'graph_info'),
origin_nodes_positions, barfem_input_nodes,
origin_input_vectors, barfem_output_nodes,
origin_output_vectors, origin_frozen_nodes,
current_edges_indices, current_edges_thickness)
# env.render(os.path.join(
# log_dir, 'render_image/{}.png'.format(epoch+1)))
if gif:
env = BarFemGym(origin_nodes_positions, barfem_input_nodes,
origin_input_vectors, barfem_output_nodes,
origin_output_vectors, origin_frozen_nodes,
current_edges_indices, current_edges_thickness,
origin_frozen_nodes)
env.reset()
env.render(os.path.join(log_dir, 'render_image/{}.png'.format(epoch)))
history['epoch'].append(epoch + 1)
history['result_efficiency'].append(current_efficiency)
# 学習履歴を保存
with open(os.path.join(log_dir, 'history.pkl'), 'wb') as f:
pickle.dump(history, f)
with open(os.path.join(log_dir, "progress.txt"), mode='a') as f:
f.writelines('epoch %d, result_efficiency: %.5f\n' %
(epoch + 1, current_efficiency))
with open(os.path.join(log_dir, "represent_value.txt"), mode='w') as f:
f.writelines('epoch %d, best_efficiency: %.5f\n' %
(best_epoch + 1, best_efficiency))
plot_efficiency_history(history,
os.path.join(log_dir, 'learning_effi_curve.png'))
error_num = 0
tmax = 10000 # 共益勾配法の最大試行回数
eps = 1e-8 # 共益勾配法の許容閾値
for i in tqdm(range(1000)):
nodes_pos, edges_indices, edges_thickness,\
input_nodes, input_vectors, frozen_nodes = \
make_random_fem_condition_with_ER(max_node_num, max_edge_pos)
output_nodes = [0]
output_vectors = np.array([[1, 1]])
result = compare_apdl_barfem(nodes_pos, edges_indices, edges_thickness,
input_nodes, input_vectors, frozen_nodes, tmax, eps)
if not result:
error_num += 1
log_dir = "check_fem_by_ansys/conditions/error{}".format(error_num)
os.makedirs(log_dir, exist_ok=True)
np.save(os.path.join(log_dir, 'nodes_pos.npy'), nodes_pos)
np.save(os.path.join(log_dir, 'edges_indices.npy'), edges_indices)
np.save(os.path.join(log_dir, 'edges_thickness.npy'), edges_thickness)
np.save(os.path.join(log_dir, 'input_nodes.npy'), input_nodes)
np.save(os.path.join(log_dir, 'input_vectors.npy'), input_vectors)
np.save(os.path.join(log_dir, 'frozen_nodes.npy'), frozen_nodes)
env = BarFemGym(nodes_pos, input_nodes, input_vectors,
output_nodes, output_vectors, frozen_nodes,
edges_indices, edges_thickness, frozen_nodes)
env.reset()
env.render(save_path=os.path.join(log_dir, "image.png"), display_number=True)
frozen_nodes)
env.reset()
efficiency = env.calculate_simulation()
#env.render(save_path="image/image_preprocess.png", display_number=True)
# 同じノード位置にあるものを排除する．
processed_nodes_pos, processed_edges_indices, processed_edges_thickness = preprocess_graph_info(
nodes_pos, edges_indices, edges_thickness)
# 傾きが一致するものをグループ分けし，エッジ分割を行う．
processed_edges_indices, processed_edges_thickness = separate_same_line_procedure(
processed_nodes_pos, processed_edges_indices, processed_edges_thickness)
processed_nodes_pos, processed_edges_indices, processed_edges_thickness =\
seperate_cross_line_procedure(processed_nodes_pos, processed_edges_indices, processed_edges_thickness)
# 同じノード，[1,1]などのエッジの排除，エッジのソートなどを行う
processed_nodes_pos, processed_edges_indices, processed_edges_thickness = \
preprocess_graph_info(processed_nodes_pos, processed_edges_indices, processed_edges_thickness)
input_nodes, output_nodes, frozen_nodes, processed_edges_thickness \
= conprocess_seperate_edge_indice_procedure(input_nodes, output_nodes, frozen_nodes, condition_nodes_pos, condition_edges_indices, condition_edges_thickness,
processed_nodes_pos, processed_edges_indices, processed_edges_thickness)
env = BarFemGym(processed_nodes_pos, input_nodes, input_vectors, output_nodes,
output_vectors, frozen_nodes, processed_edges_indices,
processed_edges_thickness, frozen_nodes)
env.reset()
efficiency = env.calculate_simulation()
env.render(save_path="image/image_proprocess.png", display_number=True)
print(efficiency)
