Validation of COSMICA code for massive stochastic simulation of cosmic rays propagation in the heliosphere

The propagation of Galactic Cosmic Rays (GCRs) within the heliosphere is governed by the Parker Transport Equation (PTE), which can be efficiently solved using a Stochastic Differential Equation (SDE)–Monte Carlo formulation. Building upon the first part of this work, where the GPU-based COSMICA code was introduced and optimized for high-performance computing, we here present its validation against a benchmark heliospheric model. COSMICA implements a three-dimensional SDE solver in CUDA/C++ optimized for multi-GPU execution, enabling the simulation of billions of quasi-particle trajectories with unprecedented efficiency. Benchmarking against the closed-source HelMod-4/CUDA model demonstrates that COSMICA achieves runtime reductions exceeding an order of magnitude while maintaining full consistency with reference fluxes. Statistical tests confirm that the distributions of output spectra from COSMICA and HelMod-4/CUDA are indistinguishable within numerical uncertainties. Furthermore, COSMICA enables the analysis of stochastic path properties – such as exit-point distributions, propagation times, and residence times – thereby providing new insights into the modulation process across varying solar activity scenarios. These results establish COSMICA as both computationally superior and scientifically reliable, making it a powerful open-source framework for heliospheric transport studies and a versatile platform for future extensions to more complex physical models.