Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Physics”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,243 records · Page 69
All sourcesNASA NTRSDOE OSTI
DOE OSTI2023

Isospin 0 and 2 two-pion scattering at physical pion mass using all-to-all propagators with periodic boundary conditions in lattice QCD

A study of two-pion scattering for the isospin channels, 𝐼 = 0 and 𝐼 = 2, using lattice QCD is presented. Möbius domain-wall fermions, on top of the Iwasaki-DSDR gauge action for gluons with periodic boundary conditions, are used for the lattice computations, which are carried out on two ensembles of gauge field configurations generated by the RBC and UKQCD Collaborations with physical masses, inverse lattice spacings of 1.023 and 1.378 GeV, and spatial extents of 𝐿 = 4.63 and 4.58 fm, respectively. The all-to-all propagator method is employed to compute a matrix of correlation functions of two-pion operators. The generalized eigenvalue problem (GEVP) is solved for a matrix of correlation functions to extract phase shifts with multiple states—two pions with a nonzero relative momentum, as well as two pions at rest. Our results for phase shifts for both the 𝐼 = 0 and 𝐼 = 2 channels are consistent with the Roy equation and chiral perturbation theory, though at this preliminary stage our errors for 𝐼 = 0 are large. An important outcome of this work is that we are successful in extracting two-pion excited states, which are useful for studying 𝐾 → 𝜋⁢𝜋 decay, on physical-mass ensembles using the GEVP.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Toward UV models of kinetic mixing and portal matter. V. Indirect probes of the new physics scale

Kinetic mixing of the dark photon, the gauge boson of a hidden U ( 1 ) D , with the Standard Model (SM) gauge fields to induce an interaction between ordinary matter and dark matter (DM) at 1-loop requires the existence of portal matter (PM) fields having both dark and SM charges. As discussed in earlier work, these same PM fields can also lead to other loop-level mechanisms besides kinetic mixing that can generate significant interactions between SM fermions and the dark photon in a manner analogous to those that can be generated between a Dirac neutrino and a SM photon, i.e., dark moments. In either case, there are reasons to believe, e.g., due to the renormalization group equation running of the U ( 1 ) D gauge coupling, that PM fields may have ∼ TeV -scale masses that lie at or above those directly accessible to the HL-LHC. If they lie above the reach of the HL-LHC, then the only way to possibly explore the physics at this high scale in the short term is via indirect measurements made at lower energies, e.g., at lepton colliders operating in the m Z to 1 TeV range. In particular, processes such as e + e − → γ + DM or e + e − → f ¯ f , where f is a SM fermion, may be most useful in this regard. Here we explore these possibilities within the framework of a simple toy PM model, introduced in earlier work, based on a non-Abelian dark gauge group completion operating at the PM scale. In the kinetic mixing setup, we show these efforts fail due to the inherently tiny cross sections in the face of substantial SM backgrounds. However, in the case of interactions via induced dark moments, since they necessarily take the form of higher dimensional operators whose influence grows with energy, we show that access to PM-scale information may become possible for certain ranges of the toy model parameters for both of these e + e − processes at a 1 TeV collider. Published by the American Physical Society 2024

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2024

Data-driven and model-agnostic approach to solving combinatorial assignment problems in searches for new physics

We present a novel approach to solving combinatorial assignment problems in particle physics. The correct assignment of decay products to parent particles is achieved in a model-agnostic fashion by introducing a neural network architecture, asswd-, which combines a custom layer based on attention mechanisms and dual autoencoders. We demonstrate how the network, trained purely on background events in an unsupervised setting, is capable of reconstructing correctly hypothetical new particles regardless of their mass, decay multiplicity, and substructure, and produces simultaneously an anomaly score that can be used to efficiently suppress the background. This model allows the extension of the suite of searches for localized excesses to include nonresonant particle pair production where the reconstruction of the two resonant masses is thwarted by combinatorics. Published by the American Physical Society 2024

Astronomy & Astrophysics↗
DOE OSTI2024

Graph neural network for neutrino physics event reconstruction

Liquid argon time projection chamber (LArTPC) detector technology offers a wealth of high-resolution information on particle interactions, and leveraging that information to its full potential requires sophisticated automated reconstruction techniques. Here, this article describes NUGRAPH 2, a graph neural network for low-level reconstruction of simulated neutrino interactions in a LArTPC detector. Simulated neutrino interactions in the MicroBooNE detector geometry are described as heterogeneous graphs, with energy depositions on each detector plane forming nodes on planar subgraphs. The network utilizes a multihead attention message-passing mechanism to perform background filtering and semantic labeling on these graph nodes, identifying those associated with the primary physics interaction with 98.0% efficiency and labeling them according to particle type with 94.9% efficiency. The network operates directly on detector observables across multiple two-dimensional representations but utilizes a three-dimensional-context-aware mechanism to encourage consistency between these representations. Model inference takes 0.12 s / event on a CPU and 0.005 s / event batched on a GPU. This architecture is designed to be a general-purpose solution for particle reconstruction in neutrino physics, with the potential for deployment across a broad range of detector technologies, and offers a core convolution engine that can be leveraged for a variety of tasks beyond the two described in this paper.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Physical-mass calculation of ρ ( 770 ) and K * ( 892 ) resonance parameters via π π and K π scattering amplitudes from lattice QCD

We present our study of the ρ ( 770 ) and K * ( 892 ) resonances from lattice quantum chromodynamics (QCD) employing domain-wall fermions at physical quark masses. We determine the finite-volume energy spectrum in various momentum frames and obtain phase-shift parametrizations via the Lüscher formalism and as a final step the complex resonance poles of the π π and K π elastic scattering amplitudes via an analytical continuation of the models. By sampling a large number of representative sets of underlying energy-level fits, we also assign a systematic uncertainty to our final results. This is a significant extension to data-driven analysis methods that have been used in lattice QCD to date, due to the two-step nature of the formalism. Our final pole positions, M + i Γ / 2 , with all statistical and systematic errors exposed, are M K * = 893 ( 2 ) ( 8 ) ( 54 ) ( 2 ) MeV and Γ K * = 51 ( 2 ) ( 11 ) ( 3 ) ( 0 ) MeV for the K * ( 892 ) resonance and M ρ = 796 ( 5 ) ( 15 ) ( 48 ) ( 2 ) MeV and Γ ρ = 192 ( 10 ) ( 28 ) ( 12 ) ( 0 ) MeV for the ρ ( 770 ) resonance. The four differently grouped sources of uncertainties are, in the order of occurrence: statistical, data-driven systematic, an estimation of systematic effects beyond our computation (dominated by the fact that we employ a single lattice spacing), and the error from the scale-setting uncertainty on our ensemble. Published by the American Physical Society 2025

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2025

Solving key challenges in collider physics with foundation models

Foundation models are neural networks that are capable of simultaneously solving many problems. Large language foundation models like ChatGPT have revolutionized many aspects of daily life, but their impact for science is not yet clear. In this paper, we use a new foundation model for hadronic jets to solve three key challenges in collider physics. In particular, we show how experiments can (1) save significant computing power when developing reconstruction algorithms, (2) perform a complete uncertainty quantification for high-dimensional measurements, and (3) search for new physics with model agnostic methods using low-level inputs. In each case, there are significant computational or methodological challenges with current methods that limit the science potential of deep learning algorithms. By solving each problem, we take jet foundation models beyond proof-of-principle studies and into the toolkit of practitioners.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2019

Direct-drive double-shell implosion: A platform for burning-plasma physics studies

Double-shell ignition designs have been investigated with the indirect-drive inertial confinement fusion (ICF) scheme in both simulations and experiments in which the inner-shell kinetic energy was limited to ~10 to 15 kJ, even driven by megajoule-class lasers such as the National Ignition Facility. Since direct-drive ICF can couple more energy to the imploding shells, we have performed a detailed study on direct-drive double-shell (D 3 S) implosions with state-of-the-art physics models implemented in radiation-hydrodynamic codes (LILAC and DRACO), including nonlocal thermal transport, cross-beam energy transfer (CBET), and first-principles-based material properties. To mitigate classical unstable interfaces, we have proposed the use of a tungsten/beryllium–mixed inner shell with gradient-density layers that can be made by magnetron sputtering. In our D 3 S designs, a 70-μm-thick beryllium outer shell is driven symmetrically by a high-adiabat (α ≥ 10), 1.9-MJ laser pulse to a peak velocity of ~240 km/s. Upon spherical impact, the outer shell transfers ~30 to 40 kJ of kinetic energy to the inner shell filled with deuterium–tritium gas or liquid, giving neutron-yield energies of ~6 MJ in 1-D simulations. Two-dimensional high-mode DRACO simulations indicated that such high-adiabat D 3 S implosions are not susceptible to laser imprint, but the long-wavelength perturbations from the laser port configuration along with CBET can be detrimental to the target performance. Yet, neutron yields of ~0.3- to 1.0-MJ energies can still be obtained from our high-mode DRACO simulations. The robust α-particle bootstrap is readily reached, which could provide a viable platform for burning-plasma physics studies. Once CBET mitigation and/or more laser energy becomes available, we believe that breakeven or moderate energy gain might be feasible with the proposed D 3 S scheme.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

Unraveling the intrinsic atomic physics behind x-ray absorption line shifts in warm dense silicon plasmas

Here, we present a novel free-energy density functional theory (DFT)-based methodology for optical property calculations of warm dense matter to cover a wide range of thermodynamic conditions and photon energies including the entire x-ray range. It uses Mermin–Kohn–Sham density functional theory with exchange-correlation (XC) thermal effects taken into account via a fully temperature dependent generalized gradient approximation XC functional. The methodology incorporates a combination of the ab initio molecular dynamics (AIMD) snapshotted Kubo–Greenwood optic data with a single-atom in simulation cell calculations to close the photon energy gap between the L and K edges and extend the K-edge tail toward many-keV photon energies. This gap arises in the standard scheme due to a prohibitively large number of bands required for the Kubo–Greenwood calculations with AIMD snapshots. Kubo–Greenwood data on snapshots provide an accurate description of optic properties at low photon frequencies slightly beyond the L-edge and x-ray absorption near edges structure (XANES) spectra, while data from periodic calculations with single atoms cover the tail regions beyond the edges. To demonstrate its applicability to mid-Z materials where the standard DFT-based approach is not computationally feasible, we have applied it to opacity calculations of warm dense silicon plasmas. These first-principles calculations revealed a very interesting phenomon of red-to-blue shift in K–L (1s → 2p) and K-edge absorptions along both isotherm and isochore, which are absent in most continuum-lowering models of traditional plasma physics. This new physics phenomenon can be attributed to the underlying competition between the screening of deeply bound core electrons and the screening of outer-shell electrons caused by warm-dense-plasma conditions. We further demonstrate that the ratio of 1s → 2p to the K-edge x-ray absorptions can be used to characterize warm-dense-plasma conditions. Eventually, based on our absorption calculations, we have established a first-principles opacity table (FPOT) for silicon in a wide range of material densities and temperatures.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

Physics-informed machine learning of the Lagrangian dynamics of velocity gradient tensor

Reduced models describing the Lagrangian dynamics of the velocity gradient tensor (VGT) in homogeneous isotropic turbulence (HIT) are developed under the physics-informed machine learning (PIML) framework. We consider the VGT at both Kolmogorov scale and coarse-grained scale within the inertial range of HIT. Building reduced models requires resolving the pressure Hessian and subfilter contributions, which is accomplished by constructing them using the integrity bases and invariants of the VGT. The developed models can be expressed using the extended tensor basis neural network (TBNN) introduced by Ling et al. [J. Fluid Mech. 807, 155 (2016)]. Physical constraints, such as Galilean invariance, rotational invariance, and incompressibility condition, are thus embedded in the models explicitly. Our PIML models are trained on the Lagrangian data from a high-Reynolds number direct numerical simulation (DNS). To validate the results, we perform a comprehensive out-of-sample test. We observe that the PIML model provides an improved representation for the magnitude and orientation of the small-scale pressure Hessian contributions. Statistics of the flow, as indicated by the joint PDF of second and third invariants of the VGT, show good agreement with the “ground-truth” DNS data. A number of other important features describing the structure of HIT are reproduced by the model successfully. We have also identified challenges in modeling inertial range dynamics, which indicates that a richer modeling strategy is required. This helps us identify important directions for future research, in particular towards including inertial range geometry into the TBNN.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Multiscale Physics of Atomic Nuclei from First Principles

Atomic nuclei exhibit multiple energy scales ranging from hundreds of MeV in binding energies to fractions of an MeV for low-lying collective excitations. As the limits of nuclear binding are approached near the neutron and proton drip lines, traditional shell structure starts to melt with an onset of deformation and an emergence of coexisting shapes. It is a long-standing challenge to describe this multiscale physics starting from nuclear forces with roots in quantum chromodynamics. Here, we achieve this within a unified and nonperturbative quantum many-body framework that captures both short- and long-range correlations starting from modern nucleon-nucleon and three-nucleon forces from chiral effective field theory. The short-range (dynamic) correlations which account for the bulk of the binding energy are included within a symmetry-breaking framework, while long-range (static) correlations (and fine details about the collective structure) are included by employing symmetry projection techniques. Our calculations accurately reproduce—within theoretical error bars—available experimental data for low-lying collective states and the electromagnetic quadrupole transitions in 20−30 Ne. In addition, we reveal coexisting spherical and deformed shapes in 30 Ne, which indicates the breakdown of the magic neutron number 𝑁 = 20 as the key nucleus 28 O is approached, and we predict that the drip line nuclei 32,34 Ne are strongly deformed and collective. By developing reduced-order models for symmetry-projected states, we perform a global sensitivity analysis and find that the subleading singlet 𝑆-wave contact and a pion-nucleon coupling strongly impact nuclear deformation in chiral effective field theory. The techniques developed in this work clarify how microscopic nuclear forces generate the multiscale physics of nuclei spanning collective phenomena as well as short-range correlations and allow one to capture emergent and dynamical phenomena in finite fermion systems such as atom clusters, molecules, and atomic nuclei.

74 ATOMIC AND MOLECULAR PHYSICS↗
DOE OSTI2020

Dynamic Microgrids in Resilient Distribution Systems With Reconfigurable Cyber-Physical Networks

Modern distribution systems energized by inverter-interfaced distributed generators (DGs) operate as coupled cyber-physical networks (C/P-networks), where the controllable components in the physical network (P-network) are coordinated through the cyber network (C-network). The concept of dynamic microgrids (MGs) operation has been adopted to enable distribution system autonomous operation with varying electric boundaries. To further enhance system operation resiliency and flexibility, dynamic MGs operation with reconfigurable C/P-networks is discussed in this work. An evaluation framework is proposed to assess the operational feasibility of distribution feeders with multiple inverter-based dynamic MGs and come out with possible restoration solutions in the context of cross-layer C/P-network reconfiguration. Furthermore, distributed controllers are developed for components with different operating characteristics to realize seamless system topology variations and provide coordinated secondary regulation in various operation modes. Furthermore, the proposed evaluation framework along with the developed distributed controller has been validated using a Hardware-in-the-Loop (HIL) real-time CPS testbed.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Enhancement of Distribution System State Estimation Using Pruned Physics-Aware Neural Networks

Realizing complete observability in the three-phase distribution system remains a challenge that hinders the implementation of classic state estimation algorithms. In this paper, a new method, called the pruned physics-aware neural network (P2N2), is developed to improve the voltage estimation accuracy in the distribution system. The method relies on the physical grid topology, which is used to design the connections between different hidden layers of a neural network model. To verify the proposed method, a numerical simulation based on one- year smart meter data of load consumptions for three-phase power flow is developed to generate the measurement and voltage state data. The IEEE 123-node system is selected as the test network to benchmark the proposed algorithm against the classic weighted least squares (WLS). Numerical results show that P2N2 outperforms WLS in terms of data redundancy and estimation accuracy.

distribution system state estimation↗
DOE OSTI2024

REBCO Insert Coils toward High-Field, Large-Bore Magnet for Quantum Physics Research

High-temperature superconducting (HTS) magnets offer a promising solution for generating high magnetic fields efficiently and economically, serving as a vital component in nextgeneration scientific instruments and carbon-neutral power systems, notably in cost-effective compact fusion reactors. These high magnetic fields play a pivotal role in advancing research on quantum materials, particularly in elucidating the intricate electronic states near magnetic phase transitions. Here, this paper delves into the research conducted at the Princeton Plasma Physics Laboratory, which is divided into three phases aimed at overcoming technical hurdles in creating large-bore, high-field HTS magnets for cutting-edge quantum physics research. In Phase 1, we designed, constructed, and tested a compact REBCO solenoidal magnet comprising six double-pancake coils featuring a 41.3-mm inner diameter and a 70.0-mm outer diameter. It adopts a no-insulation approach to ensure electrical and thermal stability. Successful testing in a saturated liquid nitrogen bath confirmed its capability to generate a 0.8-T field at 77 K and 2 T at 65 K, with ongoing integration into liquid helium testing. Building on this experience, Phase 2 involves the design of an HTS insert coil, intended to nest within a 12-T low-temperature superconducting (LTS) outsert to achieve a minimum of 20 T at 4.2 K. The focus of Phase 2 centers on addressing challenges posed by screening-current (SC) effects, particularly the associated stress/strain issues. Employing a numerical model that fully couples the SC and mechanical analysis, we discuss strategies for managing stress to mitigate the effects of SCinduced stress concentrations.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2024

Voltage Calculations in Secondary Distribution Networks via Physics-Inspired Neural Network Using Smart Meter Data

The increasing penetration of distributed energy resources (DERs) leads to voltage issues across distribution networks, necessitating voltage calculations by utilities. Electric model-free voltage calculation offers an enticing solution. However, most researches mainly focus on primary distribution networks ignoring secondary distribution networks and commonly overlook extreme voltage case calculations, which require the model’s extrapolation abilities. Here, in addressing the gaps, this paper presents a customized physics-inspired neural network (PINN) model, the structure of which is inspired by the derived coupled power flow model of primary-secondary distribution networks. To ensure precision and rapid convergence, a crafted training framework for the PINN model is proposed. The PINN’s “structure-mimetic” design enables superior extrapolation for unseen scenarios and enhances physical information awareness. We demonstrate this through two applications: hosting capacity analysis and customer-transformer connectivity. The effectiveness and advantages of the proposed PINN model are validated on two public testing systems and one utility distribution feeder model.

Distribution network↗
DOE OSTI2023

Role of reinforcement learning for risk-based robust control of cyber-physical energy systems

Critical infrastructures such as cyber-physical energy systems (CPS-E) integrate information flow and physical operations that are vulnerable to natural and targeted failures. Safe, secure, and reliable operation and control of CPS-E is critical to ensure societal well-being and economic prosperity. Automated control is key for real-time operations and may be mathematically cast as a sequential decision-making problem under uncertainty. Emergence of data-driven techniques for decision making under uncertainty, such as reinforcement learning (RL), have led to promising advances for addressing sequential decision-making problems for risk-based robust CPS-E control. However, existing research challenges include understanding the applicability of RL methods across diverse CPS-E applications, addressing the effect of risk preferences across multiple RL methods, and development of open-source domain-aware simulation environments for RL experimentation within a CPS-E context. This article systematically analyzes the applicability of four types of RL methods (model-free, model-based, hybrid model-free and model-based, and hierarchical) for risk-based robust CPS-E control. Problem features and solution stability for the RL methods are also discussed. We demonstrate and compare the performance of multiple RL methods under different risk specifications (risk-averse, risk-neutral, and risk-seeking) through the development and application of an open-source simulation environment. Motivating numerical simulation examples include representative single-zone and multizone building control use cases. Finally, six key insights for future research and broader adoption of RL methods are identified, with specific emphasis on problem features, algorithmic explainability, and solution stability.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2023

Physics-Informed Deep Learning-Based Proof-of-Concept Study of a Novel Elastohydrodynamic Seal for Supercritical CO 2 Turbomachinery

Supercritical carbon dioxide (sCO 2 ) power cycles show promising potential of higher plant efficiencies and power densities for a wide range of power generation applications such as fossil fuel power plants, nuclear power production, solar power, and geothermal power generation. sCO 2 leakage through the turbomachinery has been one of the main concerns in such applications. To offer a potential solution, we propose an elastohydrodynamic (EHD) seal that can work at elevated pressures and temperatures with low leakage and minimal wear. The EHD seal has a very simple, sleeve-like structure, wrapping on the rotor with minimal initial clearance at micron levels. Here, in this work, a proof-of-concept study for the proposed EHD seal was presented by using the simplified Reynolds equation and Lame’s formula for the fluid flow in the clearance and for seal deformation, respectively. The set of nonlinear equations was solved by using both the conventional Prediction–Correction (PC) method and modern Physics-Informed Neural Network (PINN). It was shown that the physics-informed deep learning method provided good computational efficiency in resolving the steep pressure gradient in the clearance with good accuracy. The results showed that the leakage rates increased quadratically with working pressures and reached a steady-state at high-pressure values of 15~20 MPa, where Q = 300 g/s at 20 MPa for an initial seal clearance of 255 μm. This indicates that the EHD seal could be tailored to become a potential solution to minimize the sCO 2 discharge in power plants.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2022

Generative Ensemble Regression: Learning Particle Dynamics from Observations of Ensembles with Physics-informed Deep Generative Models

Here, we propose a new method for inferring the governing stochastic ordinary differential equations (SODEs) by observing particle ensembles at discrete and sparse time instants, i.e., multiple “snapshots.” Particle coordinates at a single time instant, possibly noisy or truncated, are recorded in each snapshot but are unpaired across the snapshots. By training a physics-informed generative model that generates “fake” sample paths, we aim to fit the observed particle ensemble distributions with a curve in the probability measure space, which is induced from the inferred particle dynamics. We employ different metrics to quantify the differences between distributions, e.g., the sliced Wasserstein distances and the adversarial losses in generative adversarial networks. We refer to this method as generative “ensemble-regression” (GER), in analogy to the classic “point-regression,” where we infer the dynamics by performing regression in the Euclidean space. We illustrate the GER by learning the drift and diffusion terms of particle ensembles governed by SODEs with Brownian motions and Lévy processes up to 100 dimensions. We also discuss how to treat cases with noisy or truncated observations. Apart from systems consisting of independent particles, we also tackle nonlocal interacting particle systems with unknown interaction potential parameters by constructing a physics-informed loss function. Finally, we investigate scenarios of paired observations and discuss how to reduce the dimensionality in such cases by proving a convergence theorem that provides theoretical support.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Bias-Variance Trade-Off in Physics-Informed Neural Networks with Randomized Smoothing for High-Dimensional PDEs

Physics-Informed Neural Networks (PINNs) have triggered a paradigm shift in scientific computing, leveraging mesh-free properties and robust approximation capabilities. While proving effective for low-dimensional partial differential equations (PDEs), the computational cost of PINNs remains a hurdle in high-dimensional scenarios. This is particularly pronounced when computing high-order and high-dimensional derivatives in the physics-informed loss. Randomized Smoothing PINN (RS-PINN) introduces Gaussian noise for stochastic smoothing of the original neural net model, enabling the use of Monte Carlo methods for derivative approximation, which eliminates the need for costly automatic differentiation. Despite its computational efficiency, especially in the approximation of high-dimensional derivatives, RS-PINN introduces biases in both loss and gradients, negatively impacting convergence, especially when coupled with stochastic gradient descent (SGD) algorithms. We present a comprehensive analysis of biases in RS-PINN, attributing them to the nonlinearity of the Mean Squared Error (MSE) loss as well as the intrinsic nonlinearity of the PDE itself. We propose tailored bias correction techniques, delineating their application based on the order of PDE nonlinearity. The derivation of an unbiased RS-PINN allows for a detailed examination of its advantages and disadvantages compared to the biased version. Specifically, the biased version has a lower variance and runs faster than the unbiased version, but it is less accurate due to the bias. To optimize the bias-variance trade-off, we combine the two approaches in a hybrid method that balances the rapid convergence of the biased version with the high accuracy of the unbiased version. In addition to methodological contributions, we present an enhanced implementation of RS-PINN. Extensive experiments on diverse high-dimensional PDEs, including Fokker-Planck, Hamilton-Jacobi-Bellman (HJB), viscous Burgers’, Allen-Cahn, and Sine-Gordon equations, illustrate the bias-variance trade-off and highlight the effectiveness of the hybrid RS-PINN. Empirical guidelines are provided for selecting biased, unbiased, or hybrid versions, depending on the dimensionality and nonlinearity of the specific PDE problem.

97 MATHEMATICS AND COMPUTING↗