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 reveals that the "multiple-descent" phenomenon in LSTM training, where performance fluctuates after overtraining, is driven by repeated order-chaos phase transitions, with the model achieving optimal performance at the first critical transition point where the "edge of chaos" is widest, enabling the most effective exploration of weight configurations.
This study utilizes the RAVEL code to simulate the first 22 hours of low-velocity ejecta evolution following the DART impact on Dimorphos, revealing that rapid, asymmetric re-accretion concentrates most mass back on the surface within 5 hours while surface topography and mechanical processes shape distinct deposit patterns that offer testable predictions for the ESA Hera mission.
This paper introduces Liquid Random Feature Methods (L-RFM), a mesh-free approach for time-dependent PDEs that enhances approximation accuracy in stiff and multi-scale regimes by replacing static temporal activations with closed-form liquid time-constant responses while retaining the computational efficiency of linear least-squares solvers.
This paper presents and validates a GPU-native deep neural network surrogate that accelerates particle-based Fokker-Planck simulations of rarefied and hypersonic flows by replacing deterministic cubic closures with efficient batched computations, achieving substantial online speedups while preserving the kinetic model's macroscopic accuracy, stability, and entropy fidelity.
This paper introduces a geometry-conditioned latent surrogate model that achieves a speed-up over high-fidelity CFD simulations by encoding adaptive mesh refinement cell-density fields as a compact proxy to accurately reconstruct transient two-phase spray breakup dynamics for efficient nozzle design.
PhysGuard is a physics-preserving framework that utilizes the empirical Fisher Information Matrix to guide gradient projection during fine-tuning, effectively reducing the sim-to-real gap in neural PDE surrogates by protecting core physical representations while adapting to experimental data.
This paper establishes that scarred eigenstates in a chaotic quantum billiard act as spatial templates for operator growth by demonstrating that orthogonal eigenstates with nearly identical probability density morphologies exhibit nearly identical out-of-time-order correlator (OTOC) dynamics, thereby linking static spatial structures to the quantitative prediction of scrambling.
This study employs modern coupled-cluster methods, including pair coupled cluster doubles (pCCD) variants, to model the catalytic mechanism of molybdenum cofactor (Moco) variants with DMSO and NO substrates, revealing the critical roles of structural relaxation, environmental effects, and orbital-based quantum information in elucidating reaction energetics and bond formation.
This paper presents a scalable, complexity Fast Multipole Method Poisson solver implemented in the RAMSES code that achieves accuracy comparable to multigrid methods while offering superior suitability for isolated boundary conditions and better parallel scalability through a single upward-downward pass hierarchy.
This paper derives working equations for transition density matrices, dipole moments, and oscillator strengths within the EOM-frozen-pair coupled-cluster framework (EOM-fpCCSD and EOM-ptCCSD) using approximations that avoid solving equations and calculating left eigenvectors, demonstrating that these models yield improved excited-state properties compared to standard EOM-CCSD.