NumCosmo

About NumCosmo

Library Overview

NumCosmo is a powerful numerical cosmology library designed to facilitate calculations of cosmological observables and analysis of statistical models. It provides a comprehensive set of tools that empower researchers and developers in the field of cosmology and astrophysics.

The library is implemented in C and takes advantage of the GObject framework, enabling a clean and object-oriented approach to development. This design choice enhances the readability and maintainability of the codebase.

Key Features

  • Numerical Cosmology: NumCosmo is tailored to perform a wide range of numerical calculations related to cosmology, such as computing cosmological distances, growth functions, power spectra, and other essential observables.

  • Statistical Model Analysis: The library facilitates the analysis of statistical models commonly used in cosmology and astrophysics, enabling researchers to study various hypotheses and theoretical frameworks effectively.

  • Comprehensive Toolset: NumCosmo offers a comprehensive set of tools that cover various aspects of cosmological research, ensuring researchers have access to the necessary functionalities for their studies.

  • Language Agnostic: Thanks to the GObject introspection framework, the library automatically generates bindings for multiple programming languages, including Python, Perl, and more. This feature ensures a broad user base and promotes cross-language collaborations.

  • Extended Python Interface: In addition to the automatic language bindings provided by GObject introspection, NumCosmo has an extended Python interface numcosmo_py that offers helpful utilities and simplified interfaces. This Python interface is designed to streamline cosmological calculations and analysis for Python developers.

By combining numerical cosmology capabilities with statistical model analysis, NumCosmo empowers researchers and developers to delve into complex cosmological phenomena, advancing the understanding of the universe and its evolution.

Available observables

  • Type Ia Supernovae
  • Baryon Acoustic Oscillations
  • Cosmic Microwave Background
  • Cluster number counts
  • Galaxy clusters mass using weak-lensing
  • Hubble data H(z)
  • General cross-correlations module

Statistical tools

  • NcmFit The library’s best-fit finding capability is highly flexible and versatile, offering the option to use different backend libraries such as GSL (GNU Scientific Library) and NLOpt. This feature allows researchers to choose the most suitable optimization library based on their specific requirements and preferences, enhancing the overall efficiency and effectiveness of the best-fit finding process.
  • Fisher Fisher Matrix: The library supports the computation of the Fisher matrix, offering parameter estimation and analysis of parameter uncertainties. It includes an adaptive numerical differentiation algorithm NcmDiff, providing efficient and accurate computation of the Fisher matrix and best-fit parameters.
  • NcmLHRatio1d and NcmLHRatio2d Likelihood Ratio Test: The library enables the computation of confidence regions using the likelihood ratio test, providing a powerful tool for parameter estimation and model comparison.
  • NcmFitMC Monte Carlo – resampling and fitting.
  • NcmFitESMCMC Ensemble Sampler MCMC – Ensemble Markov Chain Monte Carlo consists in every point of the MCMC chain being a emsemble of points in the parameter space. It implements an affine invariant move method (stretch move) and an Approximate Posterior Ensemble Sampler (APES) algorithm for efficient sampling of challenging distributions. APES utilizes kernel density estimation and radial basis interpolation, leading to faster convergence and improved acceptance probability compared to traditional MCMC methods.

All the methods mentioned above generate a catalog using NcmMSetCatalog, providing a unified approach for analyzing the results. The use of a catalog also enables support for resuming algorithms from a previous crash and enhances precision by allowing extensions to computations. This catalog-based system streamlines the analysis process and ensures robustness in handling unforeseen interruptions or the need for additional computations.

These are just a few examples of the main objects and data support offered by NumCosmo. In reality, the library encompasses a wide array of additional statistical tools and cosmological computations, making it a comprehensive and versatile toolkit for researchers and developers in the field of cosmology and astrophysics.

Authors

Post Directory