Solving the Traveling Salesman Problem with DGL

Questions
#1

Hi,
I am trying to solve routing problems with network distance (not euclidian) and constraints such as Time Window and Capacity.

There are some papers around TSP and VRP but they do not implement those requirements (here are links to the paper “Attention, Learn To Solve Routing Problems” and its code in GitHub).

Can DGL support such settings easily?

If yes, what are the requirements?
Thanks!

#2

There has been an effort of applying graph neural networks to TSP problems in Benchmarking Graph Neural Networks. They have also released their code in DGL and you may want to take a look.