903 papers
Computational physics bridges the gap between abstract theory and real-world observation by using powerful computers to solve complex physical problems. This field allows scientists to simulate everything from the collision of subatomic particles to the swirling dynamics of galaxies, offering insights that traditional experiments alone cannot provide.
On Gist.Science, we continuously process every new preprint in this category from arXiv to make these breakthroughs accessible to everyone. Each entry is accompanied by both a clear, plain-language explanation and a detailed technical summary, ensuring that researchers and curious readers alike can grasp the significance of the latest findings without getting lost in dense equations.
Below are the latest papers in computational physics, curated to keep you at the forefront of this rapidly evolving discipline.
This paper introduces a novel matrix formulation for radiative transfer in coupled mixed-boundary problems that guarantees non-negative solutions and unconditional energy conservation while resolving a previously unidentified discrepancy in classical zonal methods through a single linear solve.
This paper introduces MARUT, a scalable, GPU-accelerated high-order CFD framework featuring adaptive mesh refinement and finite-rate chemistry capabilities, designed to deliver high-fidelity simulations of compressible and reacting flows across subsonic to hypersonic regimes on exascale supercomputing architectures.
This paper introduces local surrogate models that enable efficient, linear-scaling evaluation of harmonic vibrational entropy in complex multilattices by exploiting sublattice-resolved locality to replace costly global Hessian diagonalization with reusable, symmetry-respecting regression problems.
This study utilizes molecular dynamics simulations and Nudged Elastic Band calculations to characterize the migration barriers and diffusivities of point defects in 3C-SiC, revealing a mobility hierarchy that governs the competition between recombination and aggregation processes critical for stabilizing spin-active defect centers in quantum technologies.
This study uses numerical simulations to demonstrate that aerodynamic wings on a leading motorcycle generate turbulent wakes and coherent vortices that independently and significantly destabilize a pursuing motorcycle by altering lift and wheelie tendencies, suggesting that competition governing bodies should consider banning such downforce-generating appendices to improve safety.
This paper introduces a Fast Fourier Transform algorithm for the special unitary group SU(2) that leverages Euler angle discretization, two-dimensional FFTs, and recursive Jacobi polynomials to achieve significantly higher computational efficiency than direct spectral analysis methods.
This paper introduces uncertainty-aware Nudged Elastic Band and Dimer methods that incorporate anisotropic force covariance directly into the optimizer's geometric constraints to preserve saddle-search equations, demonstrating significantly improved convergence and accuracy in locating transition states compared to standard stochastic approaches.
This paper introduces a physics-inspired continuous relaxation framework that maps discrete binary variables to complex phases, leveraging an implicit regularization mechanism derived from phase dynamics to achieve superior convergence and robustness in solving NP-hard combinatorial optimization problems like QUBO, sparse coding, and planted-solution Ising models.
This paper introduces the Method of Simultaneous Solutions (MOSS), a Monte Carlo approach that concurrently solves neutron transport and heat conduction equations to reduce computational costs, while also analyzing its limitations regarding infinite variance, boundary condition handling, and approximation errors in temperature calculations.
The paper introduces Hermite-NGP, a gradient-augmented multi-resolution hash encoding that stores function values and mixed partial derivatives at grid vertices to enable fully analytic derivative evaluation and a multigrid-style curriculum, achieving significantly lower errors and faster convergence than existing neural PDE solvers.