环境问题—安装TSP求解器concorde&LKH

Posted 2022-11-26 oliveQ

Ubuntu16安装TSP求解器

• 安装pyconcorde【成功】
• 安装concorde【qsopt无法链接】
• 安装LKH3.0.7
• 其他

安装pyconcorde【成功】

• 下载GitHub项目pyconcorde：

1. 下载到调库地址:
   /home/dell/anaconda3/envs/（虚拟环境名）/lib/python3.9/site-packages

2. 或者就在当前项目文件夹下，用os和sys搭配调取，如测试文件那样

• 下载命令：可以Windows下载，但是不要解压！！
  git clone https://github.com/jvkersch/pyconcorde
  不然安装过程会出现错误：

  collect2: error: ld returned 1 exit status
  error: command 'gcc' failed with exit status 1
  

• 打开文件：
  cd pyconcorde
  [可以]预先安装好：Cython[whl]，setuptools[pip3]

• 安装：
  pip install -e . -i https://pypi.tuna.tsinghua.edu.cn/simple
  

• 【可忽略】测试代码concorde_test.py：参考
  文件夹的设置如下：
  —pyconcorde
  —TSP
  ------concorde_test.py
  ------tsplib_data

  import os
  import sys
  os.chdir(os.path.dirname(os.path.abspath(__file__)))
  sys.path.insert(0, "..")  # for pyconcorde
  from pyconcorde.concorde.tsp import TSPSolver
  import networkx as nx
  import matplotlib.pyplot as plt
  import os
  import tsplib95
  from time import time
  
  def plot(problem, tour):
      G = problem.get_graph()
      if len(problem.node_coords) == 0:
          coord = problem.display_data
      else:
          coord = 
          for node in G.nodes:
              coord[node] = problem.node_coords[node]
  
      # remove original edges
      G.remove_edges_from(list(G.edges))
  
      # add edges
      edges = []
      for i in range(len(tour) - 1):
          n1, n2 = tour[i], tour[i + 1]
          edges.append((n1, n2))
      edges.append((tour[-1], tour[0]))
      G.add_edges_from(edges)
      nx.draw_networkx(G, coord, node_size=15, with_labels=False)
      plt.savefig("test_concorde.png", bbox_inches='tight')
      plt.show()
      
  def get_example_path():
      path = os.path.dirname(os.path.abspath(__file__))
      datapath = os.path.join(path, "tsplib_data")
      return datapath
  
  def run(filename, is_plot=True):
      datapath = get_example_path()
      filepath = os.path.join(datapath, filename)
      problem = tsplib95.load(filepath)
      start = time()
      solver = TSPSolver.from_tspfile(filepath)
      solution = solver.solve()
      end = time()
      print("Time spent to solve s".format(end - start))
      print("Optimal value: ", solution.optimal_value)
      tour = solution.tour
      tour = [t + 1 for t in tour]
      plot(problem, tour)
  
  if __name__ == "__main__":
      run("berlin52.tsp")
  

安装concorde【qsopt无法链接】

官方qsopt

• 安装命令

  #!/bin/bash
  mkdir concorde
  cd concorde
  mkdir qsopt
  cd qsopt
  # ubuntu16 [(文件)直接下载就行]
  wget http://www.math.uwaterloo.ca/~bico/qsopt/beta/codes/PIC/qsopt.PIC.a
  wget http://www.math.uwaterloo.ca/~bico/qsopt/beta/codes/PIC/qsopt.h
  wget http://www.math.uwaterloo.ca/tsp/concorde/downloads/codes/src/co031219.tgz
  tar xf co031219.tgz
  # 把解码后的文件放到与qsopt相同的文件夹下
  # concorde
  # ——qsopt
  # ——BIGGUY
  # ——configure
  # ——....
  # 在concorde文件夹下（有configure文件）执行
  bash ./configure --with-qsopt=$(pwd)/qsopt
  make
  sudo cp concorde /usr/local/bin
  TSP/concorde -s 99 -k 100
  

• 检查是否可以导入（pyconcorde->concorde）【没报错即可】

  # 如果安装的是concorde
  from concorde.tsp import TSPSolver
  from concorde.tests.data_utils import get_dataset_path
  
  # 如果安装的是pyconcorde
  from pyconcorde.concorde.tsp import TSPSolver
  from pyconcorde.concorde.tests.data_utils import get_dataset_path
  

安装LKH3.0.7

用python包：lkh，调用LKH3.0.7C文件，参考

• 预先编译LKH相关文件，windows下载到Ubuntu解压

  wget http://akira.ruc.dk/~keld/research/LKH-3/LKH-3.0.7.tgz
  tar xvfz LKH-3.0.7.tgz
  cd LKH-3.0.7
  make
  sudo cp LKH /usr/local/bin
  

• 下载python包
  pip3 install lkh
  

• 测试代码：
  因为lkh库输出solution.tour,但我想要全部solution信息
  于是修改了lkh.init.py问题
  

  import requests
  import lkh
  
  problem_str = requests.get('http://vrp.atd-lab.inf.puc-rio.br/media/com_vrp/instances/A/A-n32-k5.vrp').text
  problem = lkh.LKHProblem.parse(problem_str)
  
  solver_path = '../LKH-3.0.7/LKH'
  solution = lkh.solve(solver_path, problem=problem, max_trials=100, runs=10)
  

其他

• 想看未知对象的所有属性？

  obj.__dict__

  • 比如solution
    ‘name’: ‘berlin52.7542.tour’,
    ‘comment’: ‘Found by LKH-3 [Keld Helsgaun] Thu Jun 30 03:52:12 2022’,
    ‘type’: ‘TOUR’,
    ‘dimension’: 52,
    ‘tours’: [[1, 22, 31, 18, 3, 17, 21, 42, 7, 2, 30, 23, 20, 50, 29, 16, 46, 44, 34, 35, 36, 39, 40, 37, 38, 48, 24, 5, 15, 6, 4, 25, 12, 28, 27, 26, 47, 13, 14, 52, 11, 51, 33, 43, 10, 9, 8, 41, 19, 45, 32, 49]],
    ‘_defaults’: ‘edge_weight_type’: None, ‘edge_weight_format’: None,
    ‘_wfunc’: <function StandardProblem._create_wfunc.. at 0x7fc7e9e85488>,
    ‘_special’: None

• 出现numpy.darray size changed,.....

  因为有多个版本的numpy，或者版本不对，建议重装numpy
  pip uninstall numpy，直到numpy卸载干净
  pip install numpy，重装，之后最好重启pycharm

• concorde的结果：0.04s
  

• LKH的结果：0.009s