Simulated Annealing for QUBOs

This repo contains a simulated annealing algorithm to solve QUBOs.

Code

See python/sim_anneal/ and cpp/sim_anneal/ for an implementation in Python and C++, respectively.

Requirements:

The Python build requires NumPy and Numba and was tested on Python 3.12.

The C++ build was tested with GCC 15.1 (https://gcc.gnu.org/gcc-15/).

Input:

Each problem instance requires three arguments:

Q (square symmetric matrix): Matrix representing your QUBO model.

num_restarts (integer): Number of desired restarts.

beta_schedule (vector): 1/temperate schedule for each iteration.

Note the length of beta_schedule corresponds to the number of iterations per restart.

Usage:

See python/README.md and cpp/README.md for instructions.

Examples:

See problems/tsp/ for a Traveling Salesman Problem that is referenced in both python/examples/ and cpp/examples/.