Complexity-continuum-within-Ising-formulation-of-NP-problems

The minimisation of the Ising Hamiltonian for sparse and dense interaction matrices with exact and physics-inspired algorithms.

The implemented interaction models include:

• the Sherrington-Kirkpatrick,
• the 3-regular maximum cut,
• the Mobius ladder graphs,
• the rewired Mobius ladder graphs.

The interaction strengths can be considered unweighted or can be taken from the bimodal and Gaussian distributions.

The available methods for solving the Ising model are:

• the exact commercial solver Gurobi (the free academic license is available at Gurobi website),
• the Hopfield-Tank neural networks (implemented in python and accelerated with numba).

Requirements: numpy, scipy, numba, gurobipy.

More details about simple and hard problems for the Ising optimisation could be found in the article:

Kalinin, Kirill P., and Natalia G. Berloff. "Complexity continuum within Ising formulation of NP problems." arXiv preprint arXiv:2008.00466 (2020).

Ps If you find the code useful for your studies please consider citing the above paper.