ml4co-kit 0.2.0__tar.gz → 0.2.2__tar.gz

{ml4co-kit-0.2.0/ml4co_kit.egg-info → ml4co-kit-0.2.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ml4co-kit
-Version: 0.2.0
+Version: 0.2.2
 Summary: ml4co-kit provides convenient dataset generators for the combinatorial optimization problem
 Home-page: https://github.com/Thinklab-SJTU/ML4CO-Kit
 Author: SJTU-ReThinkLab
@@ -37,17 +37,17 @@ This repository focuses on the supporting code for method development instead of
 * Data generation of various distributions;
 * Problem and solution visualization;
 * Evaluators for different problems.
+* Common post-processing algorithms
 
 | Problem |                Data                |    solver     |      Supervision      | evaluator |      visualization      |
 | :-----: | :--------------------------------: | :-----------: | :-------------------: | :-------: | :---------------------: |
-|   TSP   | Uniform, Gaussian, Cluster, TSPLIB | LKH, Concorde, Concorde-Large, GA-EAX, GA-EAX-Large | Solution, Edge Regret |     ✔     | Problem Graph, Solution |
-|   MIS   |       SATLIB, ER, BA, HK, WS       | KaMIS, Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
-|  CVRP   |    Uniform, Gaussian, VRPLIB       | PyVRP, LKH, HGS |       Solution        |     ✔     | Problem Graph, Solution |
 |  ATSP   |    SAT, HCP       | LKH |       Solution        |     ✔     | -- |
-
-###### ML4CO Organization:
-
-<img src="docs/assets/organization.jpg" alt="Organization" width="800"/>
+|  CVRP   |    Uniform, Gaussian, VRPLIB       | PyVRP, LKH, HGS |       Solution        |     ✔     | Problem Graph, Solution |
+|   MC   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MCl   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MIS   |       SATLIB, ER, BA, HK, WS       | KaMIS, Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MVC   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   TSP   | Uniform, Gaussian, Cluster, TSPLIB | LKH, Concorde, Concorde-Large, GA-EAX, GA-EAX-Large | Solution, Edge Regret |     ✔     | Problem Graph, Solution |
 
 We are still enriching the library and we welcome any contributions/ideas/suggestions from the community. A comprehensive modular framework built upon this library that integrates core ML4CO technologies is coming soon.
 
@@ -80,6 +80,7 @@ requests>=2.31.0
 async_timeout>=4.0.3
 pyvrp>=0.6.3
 cython>=3.0.8
+gurobipy>=11.0.3
 ```
 
 To ensure you have access to all functions, such as visualization, you'll need to install the following packages using `pip`:
@@ -125,7 +126,7 @@ tsp_data_lkh = TSPDataGenerator(
     num_threads=8,
     nodes_num=50,
     data_type="uniform",
-    solver="lkh",
+    solver="LKH",
     train_samples_num=16,
     val_samples_num=16,
     test_samples_num=16,
@@ -177,101 +178,49 @@ AVG         1011.043859  1032.923407 -5.535573e-01
 
 ### Visualization
 
-#### TSP
-
-```python
-from ml4co_kit.solver import TSPConcordeSolver
-from ml4co_kit.draw.tsp import draw_tsp_solution, draw_tsp_problem
-
-# use TSPConcordeSolver to solve the problem
-solver = TSPConcordeSolver(scale=1)
-solver.from_tsplib("examples/tsp/kroA150.tsp")
-solver.solve(norm="EUC_2D")
-
-# draw
-draw_tsp_problem(
-    save_path="docs/assets/kroA150_problem.png",
-    points=solver.ori_points,
-)
-draw_tsp_solution(
-    save_path="docs/assets/kroA150_solution.png",
-    points=solver.ori_points,
-    tours=solver.tours
-)
-```
-
-Visualization Results:
+#### CVRP
 
 <p>
-<img src="docs/assets/kroA150_problem.png" width="35%" alt="" />
-<img src="docs/assets/kroA150_solution.png" width="35%" alt="" />
+<img src="docs/assets/cvrp_problem.png" width="35%" alt="" />
+<img src="docs/assets/cvrp_solution.png" width="35%" alt="" />
 </p>
 
-#### MIS
+#### MCl
 
-```python
-from ml4co_kit.solver import KaMISSolver
-from ml4co_kit import draw_mis_problem, draw_mis_solution
+<p>
+<img src="docs/assets/mcl_problem.png" width="35%" alt="" />
+<img src="docs/assets/mcl_solution.png" width="35%" alt="" />
+</p>
 
-# use KaMISSolver to solve the problem
-mis_solver = KaMISSolver()
-mis_solver.solve(src="examples/mis_example")
+#### MCut
 
-# draw
-draw_mis_problem(
-    save_path="docs/assets/mis_problem.png", 
-    gpickle_path="examples/mis/instance/mis_example.gpickle"
-)
-draw_mis_solution(
-    save_path="docs/mis_solution.png",
-    gpickle_path="examples/mis/instance/mis_example.gpickle",
-    result_path="examples/mis/solution/mis_example_unweighted.result"
-)
-```
+<p>
+<img src="docs/assets/mcut_problem.png" width="35%" alt="" />
+<img src="docs/assets/mcut_solution.png" width="35%" alt="" />
+</p>
 
-Visualization Results:
+#### MIS
 
 <p>
 <img src="docs/assets/mis_problem.png" width="35%" alt="" />
 <img src="docs/assets/mis_solution.png" width="35%" alt="" />
 </p>
 
-#### CVRP
+#### MVC
 
-```python
-from ml4co_kit import CVRPPyVRPSolver
-from ml4co_kit import draw_cvrp_problem, draw_cvrp_solution
-
-# use KaMISSolver to solve the problem
-solver = CVRPPyVRPSolver(
-    depots_scale=1,
-    points_scale=1,
-    time_limit=1
-)
-solver.from_vrplib("examples/cvrp/A-n32-k5.vrp")
-solver.solve()
-
-# draw
-draw_cvrp_problem(
-    save_path="docs/assets/cvrp_problem.png",
-    depots=solver.depots[0],
-    points=solver.points[0]
-)
-draw_cvrp_solution(
-    save_path="docs/assets/cvrp_solution.png",
-    depots=solver.depots[0],
-    points=solver.points[0],
-    tour=solver.tours
-)
-```
+<p>
+<img src="docs/assets/mvc_problem.png" width="35%" alt="" />
+<img src="docs/assets/mvc_solution.png" width="35%" alt="" />
+</p>
 
-Visualization Results:
+#### TSP
 
 <p>
-<img src="docs/assets/cvrp_problem.png" width="35%" alt="" />
-<img src="docs/assets/cvrp_solution.png" width="35%" alt="" />
+<img src="docs/assets/tsp_problem.png" width="35%" alt="" />
+<img src="docs/assets/tsp_solution.png" width="35%" alt="" />
 </p>
 
+
 ### Develop ML4CO Algorithms
 
 Please refer to `ml4co_kit/learning` for the base classes that facilitate a quick establishment of a ML4CO project. You can easily build a project by inheriting the base classes and additionally implement task-specific and methodology-specific functions according to [ML4CO Organization](#ML4CO Organization:). We provide an minimalistic exmple of build a simple ML4TSP project in `docs/project_example`.

{ml4co-kit-0.2.0 → ml4co-kit-0.2.2}/README.md

@@ -12,17 +12,17 @@ This repository focuses on the supporting code for method development instead of
 * Data generation of various distributions;
 * Problem and solution visualization;
 * Evaluators for different problems.
+* Common post-processing algorithms
 
 | Problem |                Data                |    solver     |      Supervision      | evaluator |      visualization      |
 | :-----: | :--------------------------------: | :-----------: | :-------------------: | :-------: | :---------------------: |
-|   TSP   | Uniform, Gaussian, Cluster, TSPLIB | LKH, Concorde, Concorde-Large, GA-EAX, GA-EAX-Large | Solution, Edge Regret |     ✔     | Problem Graph, Solution |
-|   MIS   |       SATLIB, ER, BA, HK, WS       | KaMIS, Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
-|  CVRP   |    Uniform, Gaussian, VRPLIB       | PyVRP, LKH, HGS |       Solution        |     ✔     | Problem Graph, Solution |
 |  ATSP   |    SAT, HCP       | LKH |       Solution        |     ✔     | -- |
-
-###### ML4CO Organization:
-
-<img src="docs/assets/organization.jpg" alt="Organization" width="800"/>
+|  CVRP   |    Uniform, Gaussian, VRPLIB       | PyVRP, LKH, HGS |       Solution        |     ✔     | Problem Graph, Solution |
+|   MC   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MCl   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MIS   |       SATLIB, ER, BA, HK, WS       | KaMIS, Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   MVC   |       SATLIB, ER, BA, HK, WS       | Gurobi |       Solution        |     ✔     | Problem Graph, Solution |
+|   TSP   | Uniform, Gaussian, Cluster, TSPLIB | LKH, Concorde, Concorde-Large, GA-EAX, GA-EAX-Large | Solution, Edge Regret |     ✔     | Problem Graph, Solution |
 
 We are still enriching the library and we welcome any contributions/ideas/suggestions from the community. A comprehensive modular framework built upon this library that integrates core ML4CO technologies is coming soon.
 
@@ -55,6 +55,7 @@ requests>=2.31.0
 async_timeout>=4.0.3
 pyvrp>=0.6.3
 cython>=3.0.8
+gurobipy>=11.0.3
 ```
 
 To ensure you have access to all functions, such as visualization, you'll need to install the following packages using `pip`:
@@ -100,7 +101,7 @@ tsp_data_lkh = TSPDataGenerator(
     num_threads=8,
     nodes_num=50,
     data_type="uniform",
-    solver="lkh",
+    solver="LKH",
     train_samples_num=16,
     val_samples_num=16,
     test_samples_num=16,
@@ -152,101 +153,49 @@ AVG         1011.043859  1032.923407 -5.535573e-01
 
 ### Visualization
 
-#### TSP
-
-```python
-from ml4co_kit.solver import TSPConcordeSolver
-from ml4co_kit.draw.tsp import draw_tsp_solution, draw_tsp_problem
-
-# use TSPConcordeSolver to solve the problem
-solver = TSPConcordeSolver(scale=1)
-solver.from_tsplib("examples/tsp/kroA150.tsp")
-solver.solve(norm="EUC_2D")
-
-# draw
-draw_tsp_problem(
-    save_path="docs/assets/kroA150_problem.png",
-    points=solver.ori_points,
-)
-draw_tsp_solution(
-    save_path="docs/assets/kroA150_solution.png",
-    points=solver.ori_points,
-    tours=solver.tours
-)
-```
-
-Visualization Results:
+#### CVRP
 
 <p>
-<img src="docs/assets/kroA150_problem.png" width="35%" alt="" />
-<img src="docs/assets/kroA150_solution.png" width="35%" alt="" />
+<img src="docs/assets/cvrp_problem.png" width="35%" alt="" />
+<img src="docs/assets/cvrp_solution.png" width="35%" alt="" />
 </p>
 
-#### MIS
+#### MCl
 
-```python
-from ml4co_kit.solver import KaMISSolver
-from ml4co_kit import draw_mis_problem, draw_mis_solution
+<p>
+<img src="docs/assets/mcl_problem.png" width="35%" alt="" />
+<img src="docs/assets/mcl_solution.png" width="35%" alt="" />
+</p>
 
-# use KaMISSolver to solve the problem
-mis_solver = KaMISSolver()
-mis_solver.solve(src="examples/mis_example")
+#### MCut
 
-# draw
-draw_mis_problem(
-    save_path="docs/assets/mis_problem.png", 
-    gpickle_path="examples/mis/instance/mis_example.gpickle"
-)
-draw_mis_solution(
-    save_path="docs/mis_solution.png",
-    gpickle_path="examples/mis/instance/mis_example.gpickle",
-    result_path="examples/mis/solution/mis_example_unweighted.result"
-)
-```
+<p>
+<img src="docs/assets/mcut_problem.png" width="35%" alt="" />
+<img src="docs/assets/mcut_solution.png" width="35%" alt="" />
+</p>
 
-Visualization Results:
+#### MIS
 
 <p>
 <img src="docs/assets/mis_problem.png" width="35%" alt="" />
 <img src="docs/assets/mis_solution.png" width="35%" alt="" />
 </p>
 
-#### CVRP
+#### MVC
 
-```python
-from ml4co_kit import CVRPPyVRPSolver
-from ml4co_kit import draw_cvrp_problem, draw_cvrp_solution
-
-# use KaMISSolver to solve the problem
-solver = CVRPPyVRPSolver(
-    depots_scale=1,
-    points_scale=1,
-    time_limit=1
-)
-solver.from_vrplib("examples/cvrp/A-n32-k5.vrp")
-solver.solve()
-
-# draw
-draw_cvrp_problem(
-    save_path="docs/assets/cvrp_problem.png",
-    depots=solver.depots[0],
-    points=solver.points[0]
-)
-draw_cvrp_solution(
-    save_path="docs/assets/cvrp_solution.png",
-    depots=solver.depots[0],
-    points=solver.points[0],
-    tour=solver.tours
-)
-```
+<p>
+<img src="docs/assets/mvc_problem.png" width="35%" alt="" />
+<img src="docs/assets/mvc_solution.png" width="35%" alt="" />
+</p>
 
-Visualization Results:
+#### TSP
 
 <p>
-<img src="docs/assets/cvrp_problem.png" width="35%" alt="" />
-<img src="docs/assets/cvrp_solution.png" width="35%" alt="" />
+<img src="docs/assets/tsp_problem.png" width="35%" alt="" />
+<img src="docs/assets/tsp_solution.png" width="35%" alt="" />
 </p>
 
+
 ### Develop ML4CO Algorithms
 
 Please refer to `ml4co_kit/learning` for the base classes that facilitate a quick establishment of a ML4CO project. You can easily build a project by inheriting the base classes and additionally implement task-specific and methodology-specific functions according to [ML4CO Organization](#ML4CO Organization:). We provide an minimalistic exmple of build a simple ML4TSP project in `docs/project_example`.

ml4co-kit-0.2.2/ml4co_kit/__init__.py

@@ -0,0 +1,91 @@
+import importlib.util
+
+#######################################################
+#                       Support                       #
+#######################################################
+from .utils.type_utils import TASK_TYPE, SOLVER_TYPE, TASK_SUPPORT_SOLVER 
+
+#######################################################
+#                      Algorithm                      #
+#######################################################
+from .algorithm import atsp_greedy_decoder, atsp_2opt_local_search
+from .algorithm import (
+    tsp_greedy_decoder, tsp_insertion_decoder, tsp_mcts_decoder, tsp_mcts_local_search
+)
+
+#######################################################
+#                     Free Dataset                    #
+#######################################################
+from .data import VRPLIBOriDataset, CVRPUniformDataset
+from .data import SATLIBOriDataset
+from .data import TSPLIBOriDataset, TSPUniformDataset, TSPLIB4MLDataset, ML4TSPDataset
+
+#######################################################
+#                      Evaluator                      #
+#######################################################
+from .evaluate import ATSPEvaluator
+from .evaluate import CVRPEvaluator, CVRPUniformEvaluator
+from .evaluate import SATLIBEvaluator
+from .evaluate import TSPEvaluator, TSPLIBOriEvaluator, TSPLIB4MLEvaluator, TSPUniformEvaluator
+
+#######################################################
+#                    Data Generator                   #
+#######################################################
+from .generator import ATSPDataGenerator
+from .generator import CVRPDataGenerator
+from .generator import MClDataGenerator
+from .generator import MCutDataGenerator
+from .generator import MISDataGenerator
+from .generator import MVCDataGenerator
+from .generator import TSPDataGenerator
+
+#######################################################
+#                        Solver                       #
+#######################################################
+from .solver import ATSPSolver, ATSPLKHSolver
+from .solver import CVRPSolver, CVRPPyVRPSolver, CVRPLKHSolver, CVRPHGSSolver
+from .solver import MClSolver, MClGurobiSolver
+from .solver import MCutSolver, MCutGurobiSolver
+from .solver import MISSolver, KaMISSolver, MISGurobiSolver
+from .solver import MVCSolver, MVCGurobiSolver
+from .solver import (
+    TSPSolver, TSPLKHSolver, TSPConcordeSolver, 
+    TSPConcordeLargeSolver, TSPGAEAXSolver, TSPGAEAXLargeSolver
+)
+
+#######################################################
+#                    Utils Function                   #
+#######################################################
+from .utils import download, compress_folder, extract_archive, _get_md5
+from .utils import iterative_execution_for_file, iterative_execution, Timer
+from .utils import np_dense_to_sparse, np_sparse_to_dense, GraphData, tsplib95
+from .utils import MISGraphData, MVCGraphData, MClGraphData, MCutGraphData
+from .utils import sat_to_mis_graph, cnf_folder_to_gpickle_folder, cnf_to_gpickle
+
+#######################################################
+#           Extension Function (matplotlib)           #
+#######################################################
+found_matplotlib = importlib.util.find_spec("matplotlib")
+if found_matplotlib is not None:
+    from .draw.cvrp import draw_cvrp_problem, draw_cvrp_solution
+    from .draw.mcl import draw_mcl_problem, draw_mcl_solution
+    from .draw.mcut import draw_mcut_problem, draw_mcut_solution
+    from .draw.mis import draw_mis_problem, draw_mis_solution
+    from .draw.mvc import draw_mvc_problem, draw_mvc_solution
+    from .draw.tsp import draw_tsp_problem, draw_tsp_solution
+
+#######################################################
+#        Extension Function (pytorch_lightning)       #
+#######################################################
+found_pytorch_lightning = importlib.util.find_spec("pytorch_lightning")
+if found_pytorch_lightning is not None:
+    from .learning.env import BaseEnv
+    from .learning.model import BaseModel
+    from .learning.train import Checkpoint, Logger, Trainer
+    from .learning.utils import to_numpy, to_tensor
+    from .learning.utils import check_dim
+    from .learning.utils import points_to_distmat, sparse_points
+
+
+__version__ = "0.2.2"
+__author__ = "SJTU-ReThinkLab"

{ml4co-kit-0.2.0 → ml4co-kit-0.2.2}/ml4co_kit/algorithm/tsp/decoder/greedy.py

@@ -10,6 +10,7 @@ def tsp_greedy_decoder(heatmap: np.ndarray) -> np.ndarray:
     # check the number of dimension
     dim_2 = False
     if heatmap.ndim == 2:
+        dim_2 = True
         heatmap = np.expand_dims(heatmap, axis=0)
     if heatmap.ndim != 3:
         raise ValueError("``heatmap`` must be a 2D or 3D array.")

ml4co-kit-0.2.2/ml4co_kit/draw/mcl.py

@@ -0,0 +1,106 @@
+import numpy as np
+import networkx as nx
+import matplotlib.pyplot as plt
+from ml4co_kit.draw.utils import get_pos_layer
+from ml4co_kit.utils.graph.mcl import MClGraphData
+
+
+def draw_mcl_problem(
+    save_path: str,
+    graph_data: MClGraphData,
+    self_loop: bool = None,
+    pos_type: str = "kamada_kawai_layout",
+    figsize: tuple = (5, 5),
+    node_color: str = "darkblue",
+    node_size: int = 20,
+    edge_color: str = "darkblue",
+    edge_alpha: float = 0.5,
+    edge_width: float = 1.0,   
+):
+    # to networkx
+    nx_graph: nx.Graph = graph_data.to_networkx()
+
+    # self loop
+    if self_loop is not None:
+        self_loop_edges = list(nx.selfloop_edges(nx_graph))
+        if self_loop == True:
+            nx_graph.add_edges_from(self_loop_edges)
+        else:
+            nx_graph.remove_edges_from(self_loop_edges)
+            
+    # pos
+    pos_layer = get_pos_layer(pos_type)
+    pos = pos_layer(nx_graph)
+
+    # plt
+    figure = plt.figure(figsize=figsize)
+    figure.add_subplot(111)
+    nx.draw_networkx_nodes(
+        G=nx_graph, pos=pos, node_color=node_color, node_size=node_size
+    )
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=nx_graph.edges, 
+        alpha=edge_alpha, width=edge_width, edge_color=edge_color
+    )
+
+    # save
+    plt.savefig(save_path)
+
+
+def draw_mcl_solution(
+    save_path: str,
+    graph_data: MClGraphData,
+    self_loop: bool = None,
+    pos_type: str = "kamada_kawai_layout",
+    figsize: tuple = (5, 5),
+    sel_node_color: str = "orange",
+    unsel_node_color: str = "darkblue",
+    node_size: int = 20,
+    clique_edge_color: str = "darkblue",
+    clique_edge_alpha: float = 1.0,
+    clique_edge_width: float = 1.0,
+    other_edge_color: str = "gray",
+    other_edge_alpha: float = 0.5,
+    other_edge_width: float = 0.5,
+):
+    # Convert to networkx graph
+    nx_graph: nx.Graph = graph_data.to_networkx()
+    nodes_label = graph_data.nodes_label
+
+    # Self-loop handling
+    if self_loop is not None:
+        self_loop_edges = list(nx.selfloop_edges(nx_graph))
+        if self_loop:
+            nx_graph.add_edges_from(self_loop_edges)
+        else:
+            nx_graph.remove_edges_from(self_loop_edges)
+
+    # Get position layout for nodes
+    pos_layer = get_pos_layer(pos_type)
+    pos = pos_layer(nx_graph)
+
+    # Prepare plot
+    figure = plt.figure(figsize=figsize)
+    figure.add_subplot(111)
+    
+    # Node colors based on selection
+    colors = [unsel_node_color if bit == 0 else sel_node_color for bit in nodes_label]
+    nx.draw_networkx_nodes(G=nx_graph, pos=pos, node_color=colors, node_size=node_size)
+    
+    # Separate edges into clique and non-clique edges
+    clique_nodes = set(np.where(nodes_label == 1)[0])  # Nodes in the maximum clique
+    clique_edges = [(u, v) for u, v in nx_graph.edges if u in clique_nodes and v in clique_nodes]
+    other_edges = [(u, v) for u, v in nx_graph.edges if (u, v) not in clique_edges]
+
+    # Draw edges in different colors
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=clique_edges, alpha=clique_edge_alpha, 
+        width=clique_edge_width, edge_color=clique_edge_color
+    )
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=other_edges, alpha=other_edge_alpha, 
+        width=other_edge_width, edge_color=other_edge_color
+    )
+
+    # Save the plot
+    plt.savefig(save_path)

ml4co-kit-0.2.2/ml4co_kit/draw/mcut.py

@@ -0,0 +1,111 @@
+import numpy as np
+import networkx as nx
+import matplotlib.pyplot as plt
+from ml4co_kit.draw.utils import get_pos_layer
+from ml4co_kit.utils.graph.mcut import MCutGraphData
+
+
+def draw_mcut_problem(
+    save_path: str,
+    graph_data: MCutGraphData,
+    self_loop: bool = None,
+    pos_type: str = "kamada_kawai_layout",
+    figsize: tuple = (5, 5),
+    node_color: str = "darkblue",
+    node_size: int = 20,
+    edge_color: str = "darkblue",
+    edge_alpha: float = 0.5,
+    edge_width: float = 1.0,   
+):
+    # to networkx
+    nx_graph: nx.Graph = graph_data.to_networkx()
+
+    # self loop
+    if self_loop is not None:
+        self_loop_edges = list(nx.selfloop_edges(nx_graph))
+        if self_loop == True:
+            nx_graph.add_edges_from(self_loop_edges)
+        else:
+            nx_graph.remove_edges_from(self_loop_edges)
+            
+    # pos
+    pos_layer = get_pos_layer(pos_type)
+    pos = pos_layer(nx_graph)
+
+    # plt
+    figure = plt.figure(figsize=figsize)
+    figure.add_subplot(111)
+    nx.draw_networkx_nodes(
+        G=nx_graph, pos=pos, node_color=node_color, node_size=node_size
+    )
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=nx_graph.edges, 
+        alpha=edge_alpha, width=edge_width, edge_color=edge_color
+    )
+
+    # save
+    plt.savefig(save_path)
+
+
+def draw_mcut_solution(
+    save_path: str,
+    graph_data: MCutGraphData,
+    self_loop: bool = None,
+    pos_type: str = "kamada_kawai_layout",
+    figsize: tuple = (5, 5),
+    sel_node_color: str = "orange",
+    unsel_node_color: str = "darkblue",
+    node_size: int = 20,
+    cut_edge_color: str = "darkblue",
+    cut_edge_alpha: float = 0.5,
+    cut_edge_width: float = 1.0,
+    other_edge_color: str = "gray",
+    other_edge_alpha: float = 0.5,
+    other_edge_width: float = 1.0,
+):
+    # to networkx
+    nx_graph: nx.Graph = graph_data.to_networkx()
+    nodes_label = graph_data.nodes_label
+
+    # self loop
+    if self_loop is not None:
+        self_loop_edges = list(nx.selfloop_edges(nx_graph))
+        if self_loop == True:
+            nx_graph.add_edges_from(self_loop_edges)
+        else:
+            nx_graph.remove_edges_from(self_loop_edges)
+                
+    # pos
+    pos_layer = get_pos_layer(pos_type)
+    pos = pos_layer(nx_graph)
+    
+    # Prepare plot
+    figure = plt.figure(figsize=figsize)
+    figure.add_subplot(111)
+    
+    # Node colors based on selection
+    colors = [unsel_node_color if bit == 0 else sel_node_color for bit in nodes_label]
+    nx.draw_networkx_nodes(G=nx_graph, pos=pos, node_color=colors, node_size=node_size)
+    
+    # cut edges
+    sel_nodes = set(np.where(nodes_label == 1)[0])  # Nodes in the maximum clique
+    cut_edges = list()
+    other_edges = list()
+    for u, v in nx_graph.edges:
+        if (u in sel_nodes and v not in sel_nodes) or (v in sel_nodes and u not in sel_nodes):
+            cut_edges.append((u, v))
+        else:
+            other_edges.append((u, v))
+    
+    # plt
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=cut_edges, alpha=cut_edge_alpha, 
+        width=cut_edge_width, edge_color=cut_edge_color
+    )
+    nx.draw_networkx_edges(
+        G=nx_graph, pos=pos, edgelist=other_edges, alpha=other_edge_alpha, 
+        width=other_edge_width, edge_color=other_edge_color
+    )
+
+    # save
+    plt.savefig(save_path)