Using noise to speed convergence of simulated annealing and markov monte carlo estimations

A quantum or classical computer system for iteratively generating statistical samples from a probability density of a model or from a state of a system comprising:

• an input module that has a configuration that receives numerical data about the system;
• a noise module that has a configuration that generates random, chaotic, or other type of numerical perturbations of the received numerical data or that generates pseudo-random noise;
• a sampler module that has a configuration that iteratively generates statistical samples from a probability density of the model or from the state of the system based on the numerical perturbations or the pseudo-random noise and the input numerical data during at least one of the iterative samplings from the probability density;
• and
• a signaling module that has a configuration that signals when information derived from successive samples of the probability density differ by less than a predetermined signaling threshold or when the number of iterations reaches a predetermined number or when the length of time since commencing the iterative estimation meets or exceeds a threshold,
• wherein:
• the sampler module has a configuration that samples from a probability density of the model or state of the system using Markov chain Monte Carlo, Gibbs sampling, quantum annealing, simulated quantum annealing, or another statistical sampling, or sub-sampling method;
• the noise module has a configuration that generates random, chaotic, or other type of numerical perturbations of the input numerical data that fully or partially satisfy a noisy Markov chain Monte Carlo (N-MCMC) condition;
• and
• the sampler module has a configuration that samples from a probability density of the model or state of the system by adding, multiplying, or otherwise combining the received numerical data with the numerical perturbations.

A quantum or classical computer system for iteratively estimating the optimal configuration of a model or state of a system comprising:

• an input module that has a configuration that receives numerical data about the system;
• a noise module that has a configuration that generates random, chaotic, or other type of numerical perturbations of the received numerical data or that generates pseudo-random noise;
• an estimation module that has a configuration that iteratively estimates the optimal configuration of the model or state of the system based on the numerical perturbations or the pseudo-random noise and the input numerical data during at least one of the iterative estimates of the optimal configuration;
• a signaling module that has a configuration that signals when successive estimates of the optimal configuration or information derived from successive estimates of the optimal configuration differ by less than a predetermined signaling threshold or when the number of estimation iterations reaches a predetermined number or when the length of time since commencing the iterative estimation meets or exceeds a threshold,
• wherein:
• the estimation module has a configuration that estimates the optimal configuration of the model or state of the system using Markov chain Monte Carlo, simulated annealing, quantum annealing, simulated quantum annealing, quantum simulated annealing, or another statistical optimization or sub-optimization method;
• the noise module has a configuration that generates random, chaotic, or other type of numerical perturbations of the input numerical data that fully or partially satisfy a noisy Markov chain Monte Carlo (N-MCMC), noisy simulated annealing (N-SA), or noisy quantum annealing (N-QA) condition;
• and
• the estimation module has a configuration that estimates the optimal configuration of the system by adding, multiplying, or otherwise combining the received numerical data with the numerical perturbations.