Search arXivSearch

arXiv · 2609.03768

Computing high-order mixed derivatives in physics-informed neural networks using multi-index Bell polynomials

Abstract

Physics-informed neural networks for high-order partial differential equations require mixed input derivatives and their gradients with respect to the network parameters. Standard implementations obtain an order-$K$ derivative by repeated automatic differentiation. We instead organize the forward recursion of the multivariate Faà di Bruno formula and its explicit backpropagation over a prescribed downward-closed set of multi-indices, with the Bell-polynomial convolutions tabulated once. The forward pass carries only the derivatives the differential operator requires. The backward pass propagates gradients from losses formed from any subset of them, including nonlinear products and coupled fields. Both recursions are exact up to roundoff and avoid nested computational graphs. An independent Taylor-jet implementation, symbolic checks of the test problems, and finite differences verify derivatives and loss gradients through order seven. On one CPU core, the method evaluates 330 mixed derivatives with respect to four inputs through order seven and the corresponding loss gradient without the memory failures observed for several nested implementations. Numerical tests include third-, fifth-, and seventh-order dispersive equations, incompressible flow, and a manufactured five-field electrohydrodynamic system. The seventh-order Zakharov-Kuznetsov test in $3+1$ dimensions has a relative solution error of $6\times10^{-4}$.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Fumihiro Imoto. 2026-09-03. Computing high-order mixed derivatives in physics-informed neural networks using multi-index Bell polynomials. https://arxiv.org/abs/2609.03768

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

A subcell-refined entropy-residual-driven limiting strategy for high-order discontinuous Galerkin methods

Fine-grained, subcell-level dissipation control is essential for achieving robust high-order discontinuous Galerkin (DG) simulations of nonlinear hyperbolic systems in under-resolved regimes while preserving accuracy. This paper proposes a subcell-refined entropy-residual-driven limiting strategy for DG on Legendre-Gauss-Lobatto nodes. The limiter introduces only nearest-neighbor pairwise dissipation within each element, with closed-form coefficients that supply the minimal dissipation required to restore the element entropy inequality. The strategy is a diagonal, locally stable approximation of classical entropy-stable methods, and a generalized subcell framework reveals split-form DG and residual-distribution-based entropy correction schemes as particular choices of the limiting coefficients. For the Euler equations, a physically consistent jump operator separately models thermal and shear entropy production while preserving velocity and pressure equilibrium; a subcell refinement of the Zhang-Shu positivity limiter ensures pointwise positivity. Extensive numerical tests confirm that the scheme maintains optimal high-order accuracy, strictly enforces entropy dissipation, and significantly reduces the difficulty of a posteriori positivity-preserving procedures.

physics.comp-ph

VNS Tokamak for Medical Isotope Production

The Volumetric Neutron Source (VNS) tokamak is a proposed fusion reactor for testing components under fusion neutron irradiation, and has potential use for radioisotope production. The VNS geometry is modeled in the Serpent 2.2.2 and OpenMC 0.15.2 neutronics codes. Coupled neutron-photon simulations compared fluxes, spectra, and selected reaction rates in the blanket and vacuum vessel. Good agreement was found overall, with the largest difference found in (n, 2n) reactions. On an HPC cluster, Serpent 2 was found to have shorter computation time in coupled simulations, while OpenMC was faster in neutron only simulations. Radioisotope production yields were simulated in Serpent 2.2.2 for capsule and Cobalt plate irradiation facilities. Results indicate potential for large volume production of 99Mo, 131I, 225Ac, 177Lu, 192Ir, 64Cu, 67Cu, 161Tb, and 153Sm while 203Pb indicates lower potential. 100Mo and LEU target heating was calculated, suggesting the LEU target mass or the cooling may need adjustment. Optimized 60Co production yielded 1.2 GBq/mg and 100,000 TBq after a 3-year irradiation period. Sensitivity to plant outage for 99Mo, 131I, 177Lu, and 60Co was simulated, suggesting irradiation can be restarted for the same isotope loading and demonstrated long-lived 60Co to be robust to long plant dwell-time.

physics.comp-ph

MadVfold: accelerating NLO event generation and reducing negative weights with SIMD vectorization and GPUs

NLO simulations are essential for LHC physics analyses but are expensive, as they are not only slow but also lead to negative weights, which imply the need to simulate much larger samples of events. Folding is a powerful technique to reduce negative weights but is itself expensive. In this paper I propose ``vectorized folding'' as a new idea to speed up these calculations using SIMD and GPUs, and I present its CUDACPP-based implementation for MG5aMC in MadVfold, including its extension for unfolded NLO event generation. Preliminary results show overall speedups around 6x to 9x with folding and 3x without it. This work is based on a test-centric, LLM-assisted software development process.

physics.comp-ph