Inquire about the TSP-GLS

[taindp98]

Dear authors,
Thanks for sharing your awesome work. I'm currently exploring your project and particularly interested in the experiment on Traveling Salesman Problem (TSP) using 2-opt. I came across this evaluation script llm4ad/task/optimization/tsp_gls_2O/evaluation.py, but I'm struggling to understand how to compare the cost and running time between the found heuristic and the optimal solution using Concorde. Could you please guide me on how to achieve this? I would greatly appreciate your assistance. Thank you in advance.

[taindp98]

Dear authors,
One more question about the training and testing set of TSP using GLS. If I understand correctly, each task has a class called GetData that generates synthetic data instances for training. Once the best heuristic is found, we test it on the testing instances, including the other synthetic sets of TSP20, TSP50, TSP100, and the TSPLib instances.
Could you kindly confirm my understanding? I would be very grateful for your assistance. Thank you in advance.

[FeiLiu36]

A class called GetData that generates synthetic data instances for training

Yes, you are right. GetData is for instance generation for training. The best heuristic can be evaluated on both synthetic and TSPLib instances.

How to compare the cost and running time between the found heuristic and the optimal solution using Concorde

For running time, we usually set a maximum running time for Concord and set a fixed maximum number of iterations for generated heuristics (i.e., GLS). The performance of heuristics is calculated as the gap to baseline (e.g., Concord or LKH).