Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BrahMap: A scalable and modular map-making framework for the CMB experiments

Avinash Anand, Giuseppe Puglisi

TL;DR

The paper tackles the challenge of scalable CMB map-making for future experiments with $10^4$–$10^5$ detectors by formulating the problem as a Generalized Least Squares optimization, where $d = P s + n$ and the GLS solution is $\hat{s} = (P^T N^{-1} P)^{-1} P^T N^{-1} d$. It introduces BrahMap, a modular framework built around linear operators that represent the pointing matrix and noise structures, with a Python interface and a C++ backend that performs fast matrix-vector products using MPI and OpenMP. The authors demonstrate substantial performance gains (e.g., $\sim 22\times$ speedup over a Python baseline) and good strong/weak scaling on CPU hardware, validating the approach for large-scale data. They also outline future extensions, including circulant noise covariance for FFT-based inverses and GPU offloading via CuPy, to further enhance scalability and applicability to realistic instrumental systematics. Overall, BrahMap provides a flexible, high-performance pathway to deliver timely sky maps for next-generation CMB experiments.

Abstract

The cosmic microwave background (CMB) experiments have reached an era of unprecedented precision and complexity. Aiming to detect the primordial B-mode polarization signal, these experiments will soon be equipped with $10^{4}$ to $10^{5}$ detectors. Consequently, future CMB missions will face the substantial challenge of efficiently processing vast amounts of raw data to produce the initial scientific outputs - the sky maps - within a reasonable time frame and with available computational resources. To address this, we introduce BrahMap, a new map-making framework that will be scalable across both CPU and GPU platforms. Implemented in C++ with a user-friendly Python interface for handling sparse linear systems, BrahMap employs advanced numerical analysis and high-performance computing techniques to maximize the use of super-computing infrastructure. This work features an overview of the BrahMap's capabilities and preliminary performance scaling results, with application to a generic CMB polarization experiment.

BrahMap: A scalable and modular map-making framework for the CMB experiments

TL;DR

The paper tackles the challenge of scalable CMB map-making for future experiments with – detectors by formulating the problem as a Generalized Least Squares optimization, where and the GLS solution is . It introduces BrahMap, a modular framework built around linear operators that represent the pointing matrix and noise structures, with a Python interface and a C++ backend that performs fast matrix-vector products using MPI and OpenMP. The authors demonstrate substantial performance gains (e.g., speedup over a Python baseline) and good strong/weak scaling on CPU hardware, validating the approach for large-scale data. They also outline future extensions, including circulant noise covariance for FFT-based inverses and GPU offloading via CuPy, to further enhance scalability and applicability to realistic instrumental systematics. Overall, BrahMap provides a flexible, high-performance pathway to deliver timely sky maps for next-generation CMB experiments.

Abstract

The cosmic microwave background (CMB) experiments have reached an era of unprecedented precision and complexity. Aiming to detect the primordial B-mode polarization signal, these experiments will soon be equipped with to detectors. Consequently, future CMB missions will face the substantial challenge of efficiently processing vast amounts of raw data to produce the initial scientific outputs - the sky maps - within a reasonable time frame and with available computational resources. To address this, we introduce BrahMap, a new map-making framework that will be scalable across both CPU and GPU platforms. Implemented in C++ with a user-friendly Python interface for handling sparse linear systems, BrahMap employs advanced numerical analysis and high-performance computing techniques to maximize the use of super-computing infrastructure. This work features an overview of the BrahMap's capabilities and preliminary performance scaling results, with application to a generic CMB polarization experiment.

Paper Structure

This paper contains 7 sections, 9 equations, 2 figures.

Table of Contents

  1. Introduction to CMB map-making
  2. Data model and formalism
  3. Map-making with BrahMap
  4. Programming implementation
  5. Performance scaling
  6. Future extensions to BrahMap
  7. Conclusion

Figures (2)

  • Figure 1: Strong scaling performance. It shows the performance speedup when the number of MPI processes is increased for the problem of fixed size.
  • Figure 2: Weak scaling performance. It shows the efficiency of parallelization as the size of problem is increased along with the number of MPI processes used in parallelization.