ana_cont

Official Resources

• Homepage: https://github.com/josefkaufmann/ana_cont
• Documentation: https://josefkaufmann.github.io/ana_cont/
• Source Repository: https://github.com/josefkaufmann/ana_cont
• License: MIT License
• Installation: pip install ana_cont

Overview

ana_cont is a Python package dedicated to the analytic continuation of many-body Green's functions. It provides a user-friendly interface to standard continuation methods, specifically Padé approximants and the Maximum Entropy Method (MaxEnt), enabling the extraction of real-frequency spectral functions from imaginary-axis QMC data.

Scientific domain: Analytic Continuation, Many-body Physics Target user community: Researchers converting Euclidean QMC data to real-frequency spectra

Theoretical Methods

• Maximum Entropy Method (MaxEnt): Probabilistic approach using Bayesian inference to find the most probable spectrum.
• Padé Approximants: Rational polynomial interpolation for continuation.
• Analytic Continuation: Inversion of $G(i\omega_n) = \int \frac{A(\omega)}{i\omega_n - \omega} d\omega$.

Capabilities (CRITICAL)

• MaxEnt Solver: Robust MaxEnt implementation (MaxEntSolver class) for spectral reconstruction.
• Padé Solver: Fast rational approximation for lower-noise data (PadeSolver class).
• Kernel Support: Supports Fermionic and Bosonic kernels (Temperature dependence).
• Preprocessing: Tools for handling input data formats and preblur.

Key Features

Python Interface:

• Easy to use Python API (OOP design).
• Integrates seamlessly with NumPy/SciPy arrays.
• Scriptable for batch processing of many self-energies.

Reliability:

• Implements standard, well-tested algorithms (Bryan's method / Historic MaxEnt).
• Configurable parameters (alpha, blur, error tolerance) for MaxEnt.
• GUI available for interactive parameter tuning.

Inputs & Outputs

• Inputs:
  • Matsubara Green's functions $G(i\omega_n)$ (complex numpy array).
  • Error estimates (scalar or covariance).
  • Real frequency grid $\omega$.
• Outputs:
  • Real-frequency Spectral Function $A(\omega)$.
  • Trace of fit quality ($\chi^2$).

Workflow and Usage

Standard post-processing tool.

1. Load $G(i\omega_n)$ from a file (e.g., TRIQS or w2dynamics output).
2. Instantiate MaxEntSolver or PadeSolver.
3. Run the solver (e.g., solver.solve()).
4. Plot the resulting spectrum.

Comparison with Other Methods

Verification & Sources

Primary sources:

1. GitHub Repository: https://github.com/josefkaufmann/ana_cont
2. Documentation: https://josefkaufmann.github.io/ana_cont/
3. Dependencies: numpy, scipy, matplotlib.

Verification status: ✅ VERIFIED

• Source code: OPEN (GitHub)
• Utility: Widely used Python tool for continuation