ZMCintegral-v5: Support for Integrations with the Scanning of Large Parameter Grids on Multi-GPUs

TL;DR

ZMCintegral-v5 introduces GPU-accelerated integration with large parameter grid scanning using Monte Carlo methods, maintaining user flexibility and optimized performance across multiple nodes.

Contribution

This version adds distributed GPU support for large-scale parameter scans with Monte Carlo integration, enhancing speed and scalability.

Findings

Effective handling of up to 10^{10} parameter points.

Maintains Python API compatibility.

Demonstrates performance on multi-node setups.

Abstract

In this updated vesion of ZMCintegral, we have added the functionality of integrations with parameter scan on distributed Graphics Processing Units(GPUs). Given a large parameter grid (up to 10^{10} parameter points to be scanned), the code will evaluate integrations for each parameter grid value. To ensure the evaluation speed, this new functionality employs a direct Monte Carlo method for the integraion. The Python API is kept the same as the previous ones and users have a full flexibility to define their own integrands. The performance of this new functionality is tested for both one node and multi-nodes conditions.