update README.md, KaMIS, and docs

README.md

@@ -102,45 +102,6 @@ matplotlib
 pytorch_lightning
 ```
 
-## Installation
-
-You can install the stable release on PyPI:
-
-```bash
-$ pip install ml4co_kit
-```
-
-or get the latest version by running:
-
-```bash
-$ pip install -U https://github.com/Thinklab-SJTU/ML4CO-Kit/archive/master.zip # with --user for user install (no root)
-```
-
-The following packages are required and shall be automatically installed by ``pip``:
-
-```
-Python>=3.8
-numpy>=1.24.4
-networkx>=2.8.8
-tqdm>=4.66.1
-pulp>=2.8.0, 
-pandas>=2.0.0,
-scipy>=1.10.1
-aiohttp>=3.9.3
-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`:
-
-```
-matplotlib
-pytorch_lightning
-```
-
 ## Usage Examples
 
 ### Solve with Traditional Solver Baselines
@@ -154,14 +115,14 @@ We provide base classes with a user-friendly approach for implementing tradition
 >>> tsp_lkh_solver = TSPLKHSolver(lkh_max_trials=500)
 
 # input instances and reference solutions by a .txt file
->>> tsp_lkh_solver.from_txt("path/to/read/tsp500_concorde.txt")
+>>> tsp_lkh_solver.from_txt("path/to/read/tsp500_concorde.txt", ref=True)
 
 # lkh solving
->>> tsp_lkh_solver.solve()
+>>> tsp_lkh_solver.solve(num_threads=10, show_time=True)
 
 # evaluate
 >>> tsp_lkh_solver.evaluate(calculate_gap=True)
-(16.583557978549532, 0.21424058722308548, 0.09031979488795724)
+(16.546304871926175, 16.545805334644392, 0.0030213676759083515, 0.009747905769875538)
 
 # save solving results
 >>> tsp_lkh_solver.to_txt("path/to/write/tsp500_lkh.txt")
@@ -177,7 +138,7 @@ 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,
@@ -227,6 +188,24 @@ gr666       3843.137961  3952.535702 -2.767786e+00
 AVG         1011.043859  1032.923407 -5.535573e-01
 ```
 
+### Algorithm
+
+```python
+>>> from ml4co-kit import TSPSolver, tsp_insertion_decoder
+
+# create solver and load data
+>>> solver = TSPSolver()
+>>> solver.from_txt("your/txt/data/path", ref=True)
+
+# use insertion algorithm to solve the problems
+>>> tours = tsp_insertion_decoder(points=solver.points)
+>>> solver.from_data(tours=tours, ref=False)
+
+# evaluate (average length, ground truth, gap, std)
+>>> solver.evaluate(calculate_gap=True)
+(6.299320133465173, 5.790133693543183, 8.816004478345556, 3.605743337834312)
+```
+
 ### Visualization
 
 Below, we use TSP, MIS, and CVRP as representative examples for illustration.
@@ -327,6 +306,6 @@ Visualization Results:
 <img src="docs/assets/cvrp_solution.png" width="35%" alt="" />
 </p>
 
-### Develop ML4CO Algorithms
+### Develop ML4CO Methods
 
-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`.
+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:).

ml4co_kit/solver/atsp/base.py

@@ -1,5 +1,6 @@
 r"""
 Basic solver for Asymmetric Traveling Salesman Problem (ATSP). 
+
 The ATSP problem requires finding the shortest tour that visits each 
 vertex of the graph exactly once and returns to the starting node 
 with the consideration that the travel costs between any two vertices 

ml4co_kit/solver/cvrp/base.py

@@ -1,5 +1,6 @@
 r"""
 Basic solver for Capacitated Vehicle Routing Problem (CVRP). 
+
 The CVRP problems requires finding the most efficient routes for a fleet of vehicles
 with limited capacity to deliver goods to a set of customers while minimizing costs.
 """
@@ -46,23 +47,23 @@ class CVRPSolver(SolverBase):
 
     :param solver_type: string, the type of the solver or algorithm used in the class.
     :param depots_scale: int, the scale of the depots.
-    :parm points_scale: int, the scale of the customer points.
-    :parm demands_scale: int, the scale of the demands of customer points.
-    :parm capacities_scale: int, the scale of the capacities of the car.
-    :parm ori_depots: np.ndarray, the original depots coordinates data read.
-    :parm depots: np.ndarray, the depots coordinates data called by the solver during solving,
+    :param points_scale: int, the scale of the customer points.
+    :param demands_scale: int, the scale of the demands of customer points.
+    :param capacities_scale: int, the scale of the capacities of the car.
+    :param ori_depots: np.ndarray, the original depots coordinates data read.
+    :param depots: np.ndarray, the depots coordinates data called by the solver during solving,
         they may initially be same as ``ori_depots``, but may later undergo standardization
         or scaling processing.
-    :parm ori_points: np.ndarray, the original customer points coordinates data read.
-    :parm points: np.ndarray, the customer points coordinates data called by the solver
+    :param ori_points: np.ndarray, the original customer points coordinates data read.
+    :param points: np.ndarray, the customer points coordinates data called by the solver
         during solving, they may initially be same as ``ori_depots``, but may later undergo
         standardization or scaling processing.
-    :parm demands: np.ndarray, the demands of each customer points.
-    :parm capacities: np.ndarray, the capacities of the car.
-    :parm tours: np.ndarray, the solution to the problems.
-    :parm ref_tours: np.ndarray, the reference solutions to the problems.
-    :parm nodes_num: int, the number of points, i.e. the sum of depots points and customer points.  
-    :parm norm: str, coordinate type. It can be a 2D Euler distance or geographic data type.
+    :param demands: np.ndarray, the demands of each customer points.
+    :param capacities: np.ndarray, the capacities of the car.
+    :param tours: np.ndarray, the solution to the problems.
+    :param ref_tours: np.ndarray, the reference solutions to the problems.
+    :param nodes_num: int, the number of points, i.e. the sum of depots points and customer points.  
+    :param norm: str, coordinate type. It can be a 2D Euler distance or geographic data type.
     """
     def __init__(
         self, 
@@ -1272,14 +1273,14 @@ def solve(
         """
         This method will be implemented in subclasses.
         
-        :parm depots: np.ndarray, the depots coordinates data called by the solver during solving,
+        :param depots: np.ndarray, the depots coordinates data called by the solver during solving,
             they may initially be same as ``ori_depots``, but may later undergo standardization
             or scaling processing.
-        :parm points:  np.ndarray, the customer points coordinates data called by the solver
+        :param points:  np.ndarray, the customer points coordinates data called by the solver
             during solving, they may initially be same as ``ori_depots``, but may later undergo
             standardization or scaling processing.
-        :parm demands: np.ndarray, the demands of each customer points.
-        :parm capacities: np.ndarray, the capacities of the car.
+        :param demands: np.ndarray, the demands of each customer points.
+        :param capacities: np.ndarray, the capacities of the car.
         :param norm: boolean, the normalization type for node coordinates.
         :param normalize: boolean, whether to normalize node coordinates.
         :param num_threads: int, number of threads(could also be processes) used in parallel.

ml4co_kit/solver/mis/kamis-source/extern/KaHIP/lib/data_structure/graph_access.h

@@ -435,6 +435,9 @@ inline void graph_access::set_partition_count(PartitionID count) {
 }
 
 inline int graph_access::build_from_metis(int n, int* xadj, int* adjncy) {
+        if(graphref != NULL) {
+                delete graphref;
+        }
         graphref = new basicGraph();
         start_construction(n, xadj[n]);
 
@@ -455,6 +458,9 @@ inline int graph_access::build_from_metis(int n, int* xadj, int* adjncy) {
 }
 
 inline int graph_access::build_from_metis_weighted(int n, int* xadj, int* adjncy, int * vwgt, int* adjwgt) {
+        if(graphref != NULL) {
+                delete graphref;
+        }
         graphref = new basicGraph();
         start_construction(n, xadj[n]);