tsp-c 0.0.25

A wrapper for c++ to solve the Traveling Salesman Problem

Package description

A wrapper for c++ to solve the Traveling Salesman Problem

Installation

pip install tsp-c

Available methods

• Greedy
• Simulated Annealing (SA)
• Particle Swarm Optimization (PSO)
• Ant Colony Optimization (ACO)
• Two-opt Heuristic

Available functions

solve_greedy(distance_matrix)
solve_SA(distance_matrix)
set_param_SA(C0, Cmin, L0, alpha)
solve_PSO(distance_matrix)
set_param_ACO(hconst, alpha, beta, evprate, intsty, nAnt, nItr)
solve_ACO(distance_matrix)
solve_two_opt(distance_matrix)
set_param_PSO(w, c1, c2, maxV, muProb, nPar, nItr)

Examples

To solve the problem with the Greedy method:

import tsp_c as tsp

distance_matrix = [
	[0.0, 290.7, 254.9, 172.9],
	[263.6, 0.0, 508.3, 185.0],
	[258.2, 497.1, 0.0, 405.6],
	[136.8, 190.7, 394.8, 0.0]
]

sol, distance = tsp.solve_greedy(distance_matrix)
print("\nSolution from Greedy:")
print(distance, " ", sol)

The solution will be (1, 3, 0, 2) with total distance 1073.8000000000002.

To solve the problem with the Simulated Annealing method, change the code to:

sol, distance = tsp.solve_SA(distance_matrix)

To set the parameters of the Simulated Annealing method, use:

tsp.set_param_SA(C0, Cmin, L0, alpha)

where

• C0 = Initial temperature (default = 20.0)
• Cmin = Final temperature (default = 0.1)
• L0 = Number of iterations in each temperature (default = 10000)
• alpha = cooling rate (default = 0.9)

For example:

tsp.set_param_SA(10.0, 0.01, 10000, 0.95)

To set the parameters of the ACO method, use:

tsp.set_param_ACO(hconst, alpha, beta, evprate, intsty, nAnt, nItr)

where

• hconst = Heuristic Constant (default = 50.0)
• alpha = Pheromone influence (default = 0.1)
• beta = Heuristic influence (default = 10)
• evprate = Evaporation Rate (default = 0.1)
• intsty = Pheromone intensity (default = 0.1)
• nAnt = Number of ants (default = 100)
• nItr = Number of iterations (default = 100)

For example:

tsp.set_param_ACO(100.0, 0.15, 4, 0.1, 0.15, 200, 200)

To set the parameters of the PSO method, use:

tsp.set_param_PSO(w, c1, c2, maxV, muProb, nPar, nItr)

where

• w = Inertia weight of the particle (default = 1.0)
• c1 = acceleration coefficient for the local best position (default = 2.0)
• c2 = acceleration coefficient for the global best position (default = 2.0)
• maxV = Maximum velocity of the particles (default = 2.0)
• muProb = mutation probability that change the position of a particle (default = 0.05)
• nPar = Number of particles (default = 100)
• nItr = Number of iterations (default = 5000)

For example:

tsp.set_param_PSO(0.8, 1.5, 1.5, 5, 0.1, 200, 1000)

Requirement

python >=3.6

Operating System

Linux

References

• Janjarassuk, Udom. (2011). A comparison of heuristic approaches to the travelling salesman problem. ICIC Express Letters. 5. 279-283.