Noise enhanced neural networks for analytic continuation

  title={Noise enhanced neural networks for analytic continuation},
  author={Juan Yao and Ce Wang and Zhi yuan Yao and Hui Zhai},
  journal={Machine Learning: Science and Technology},
  • Juan YaoCe Wang Hui Zhai
  • Published 24 November 2021
  • Computer Science
  • Machine Learning: Science and Technology
Analytic continuation maps imaginary-time Green’s functions obtained by various theoretical/numerical methods to real-time response functions that can be directly compared with experiments. Analytic continuation is an important bridge between many-body theories and experiments but is also a challenging problem because such mappings are ill-conditioned. In this work, we develop a neural network (NN)-based method for this problem. The training data is generated either using synthetic Gaussian… 



Artificial Neural Network Approach to the Analytic Continuation Problem.

This work presents a general framework for building an artificial neural network (ANN) that solves the analytic continuation problem with a supervised learning approach and shows that the method can reach the same level of accuracy for low-noise input data, while performing significantly better when the noise strength increases.

Bayesian parametric analytic continuation of Green's functions

The capability of BPAC in terms of CTQMC data for an Anderson impurity model (AIM) that shows a generalized Kondo scenario is demonstrated and its application to an AIM arising from the ab-initio treatment of SrVO$_3$.

Constrained sampling method for analytic continuation.

A method for analytic continuation of imaginary-time correlation functions to real-frequency spectral functions is proposed, and very good agreement is found with Bethe ansatz results in the ground state and with exact diagonalization of small systems at elevated temperatures.

Average spectrum method for analytic continuation: Efficient blocked-mode sampling and dependence on the discretization grid

This work introduces an efficient implementation of the average spectrum method, based on the singular value decomposition of the integral kernel, that eliminates the problem of the discretization of the real-frequency grid, which biases the results.

Numerical analytic continuation: Answers to well-posed questions

We formulate the problem of numerical analytic continuation in a way that lets us draw meaningful conclusions about the properties of the spectral function based solely on the input data. Apart from

Algorithms for optimized maximum entropy and diagnostic tools for analytic continuation.

This work analyzes all aspects of the implementation that are critical for accuracy and speed and presents a highly optimized approach to maximum entropy, which covers most typical cases for fermions, bosons, and response functions.

Extending the average spectrum method: Grid point sampling and density averaging

This paper demonstrates that sampling the grid points, instead of keeping them fixed, also changes the functional integral limit and rather helps to overcome the bias considerably, and provides an efficient algorithm for doing the sampling and shows how the density of thegrid points acts now as a default model with a significantly reduced biasing effect.

Statistical and computational intelligence approach to analytic continuation in Quantum Monte Carlo

The Genetic Inversion via Falsification of Theories method is presented, which allowed us to compute dynamical properties of strongly interacting quantum many–body systems with very high accuracy and provides a very general framework to face analytic continuation problems that could emerge in several areas of applied Science.

Analytic continuation via domain knowledge free machine learning

The machine-learning-based approach to analytic continuation not only provides the more accurate spectrum than the conventional methods in terms of peak positions and heights, but is also more robust against the noise which is the required key feature for any continuation technique to be successful.

Maximum entropy formalism for the analytic continuation of matrix-valued Green's functions

A generalization of the maximum entropy method to the analytic continuation of matrix-valued Green's functions, where all matrix elements of the Green's function matrix can be analytically continued; a computationally cheap element-wise method is introduced for this purpose.