vineetkotariya / quantum-multibody-rbm

Using Artificial Neural Networks (ANNs) to solve certain quantum multi-body problems. Based on the work by Carleo Troyer 2017

quantum-multibody-RBM

Using RBMs to solve certain quantum multi-body problems. Based on work by Carleo Troyer 2017

• Stochastic Reconfiguration used for optimizing the network
• Metropolis Hastings algorithm used for sampling (MCMC)
• Restricted Boltzmann Machine used for representing the network
• Ground state for the Ising 1D spin model is being calculated

Parameters

n_spins : Number of spins (input states)
alpha : n_hidden = n_spins * alpha
h : h_x in the Hamiltonian
n_sweeps : No. of sweeps per epoch
sweep_factor : one sweep consists of (n_spins * sweep_factor) steps (flipping each spin an expected number of n_flips times)
therm_factor : (therm_factor * n_sweeps) no. of markov process steps are discarded (for thermalization)