Search NASA⌕ Search

SEARCH · Search NASA

Results for “Discretization error”

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 685 records · Page 38

Using Directional Diffusion Coefficients for Nonlinear Diffusion Acceleration of the First Order SN Equations in Near-Void Regions

The common definition of the diffusion coeffcient as the inverse of three times the transport cross section is not compat- ible with voids. Morel introduced a non-local tensor diffusion coeffcient that remains finite in voids[1]. It can be obtained by solving an auxiliary transport problem without scattering or fission. Larsen and Trahan successfully applied this diffusion coeffcient for enhancing the accuracy of diffusion solutions of very high temperature reactor (VHTR) problems that feature large, optically thin channels in the z-direction [2]. It is demonstrated that a significant reduction of error can be achieved in particular in the optically thin region. Along the same line of thought, non-local diffusion tensors are applied modeling the TREAT reactor confirming the findings of Larsen and Trahan [3]. Previous work of the authors have introduced a flexible Nonlinear-Diffusion Acceleration (NDA) method for the first order S N equations discretized with the discontinuous finite element method (DFEM), [4], [5], [6]. This NDA method uses a scalar diffusion coeffcient in the low-order system that is obtained as the flux weighted average of the inverse transport cross section. Hence, it su?ers from very large and potentially unbounded diffusion coeffcients in the low order problem. However, it was noted that the choice of the diffusion coeffcient does not influence consistency of the method at convergence and hence the di?usion coeffcient is essentially a free parameter. The choice of the di?usion coeffcient does, however, affect the convergence behavior of the nonlinear di?usion iterations. Within this work we use Morel’s non-local di?usion coef- ficient in the aforementioned NDA formulation in lieu of the flux weighted inverse of three times the transport cross section. The goal of this paper is to demonstrate that significant en- hancement of the spectral properties of NDA can be achieved in near void regions. For testing the spectral properties of the NDA with non-local diffusion coeffcients, the periodical horizontal interface problem is used [7]. This problem consists of alternating stripes of optically thin and thick materials both of which feature scattering ratios close to unity.

97 MATHEMATICS AND COMPUTING↗

Predicting the dark matter-baryon abundance ratio

We discuss relaxation solutions to the dark matter-baryon coincidence problem in the context of QCD axion dark matter. In relaxation solutions, a moduli dynamically adjusts the mass of dark matter and baryons until their energy densities are 𝒪⁡(1) the same. Because the QCD axion is heavily connected to QCD, scanning the QCD axion mass inherently also scans the proton mass. In the context of relaxation solutions, this implies that the ratio of dark matter to baryon abundances (Ω DM /Ω B ) is a ratio of beta functions showing that these models can only accommodate discrete values of ΩDM/ΩB thereby “predicting” the ratio of the dark matter to baryon abundances. The original composite axion model has only a single integer degree of freedom 𝑁, the size of the gauge group, and we show that when 𝑁 =8 the observed value of Ω DM /Ω B =5.36 is reproduced to within its percent level error bars. Novel tests of this model include more precise measurements of Ω DM /Ω B , a better lattice determination of the dependence of the proton mass on the high energy QCD gauge coupling, as well as more traditional tests such as fifth force experiments.

79 ASTRONOMY AND ASTROPHYSICS↗

A dynamic 2D Borehole Thermal Energy Storage (BTES) model for enhanced computational efficiency

Progressing toward a future increasingly reliant on renewable energy sources, the development of effective, durable energy storage solutions becomes essential to balance supply and demand fluctuations. Borehole Thermal Energy Storage (BTES) is a long-duration thermal energy storage technology that captures excess heat generated from renewable energy sources and stores it underground for later use, enabling the efficient utilization of sustainable energy. This approach is particularly valuable in district energy networks when integrated with Ground Source Heat Pumps (GSHP) to provide stable heating and cooling. However, traditional three-dimensional (3D) numerical models of BTES systems demand extensive computational resources, limiting their practicality for real-time and large-scale applications. This study introduces a novel two-dimensional (2D) modeling approach that reduces computational costs while maintaining high accuracy. By employing a radial ring-based discretization method, the model simulates heat injection, retention, and retrieval dynamics over seasonal cycles. A new thermal-mass weighted-average temperature parameter is introduced to evaluate the performance of BTES systems. Model validation against FEFLOW simulations demonstrates a 17-fold improvement in computational speed compared to traditional Computational Fluid Dynamics (CFD) models while achieving a mean absolute percentage error (MAPE) of 2 % during charging and 4 % during discharging. Additionally, a trade-off analysis between computational efficiency and accuracy is conducted, ensuring the model's applicability for real-world scenarios. The findings of this research contribute to the development of computationally efficient BTES models, facilitating better optimization, control, and integration into renewable energy systems. This work provides a foundation for further studies in techno-economic analysis, multi-year performance evaluation, and real-time operational strategies for BTES applications, supporting a more sustainable energy future.

2D modeling↗

A hybrid embedded discrete fracture model for simulating tight porous media with complex fracture systems

Numerical simulation considering complex fracture networks in tight oil reservoir becomes a hot issue during these years. Recently, embedded discrete fracture model (EDFM) shows great advantages over the DFM approaches in both calculation efficiency and simplicity of simulation workflow. EDFM has been widely used in tight oil/gas natural energy or enhanced oil recovery (EOR) development. If a large number of small-scale natural fractures exist in reservoir, the calculation efficiency of EDFM method may decrease. A new hybrid simulation model combining the EDFM and dual porosity (DP) model is proposed. The large-scale fractures are dealt with EDFM method while small scale fractures are dealt with DP model. Considering the low permeability of matrix, transient transfer between matrix and natural fracture is considered. The proposed hybrid model is validated compared with the local grid refinement (LGR) model and EDFM-MINC model, and the difference between transient transfer and pseudo-steady-state transfer effect on production is presented. Finally, two test cases are presented, and the sensitivity of some key parameters are investigated. Result shows that the new simulation model has higher calculation efficiency than LGR and EDFM-MINC models. Meanwhile, it honors a good accuracy. The pseudo-steady-state transfer between matrix and natural fracture leads to big error in tight oil production performance analysis both in single phase and two phases flow. Low matrix permeability and big matrix size increase the production difference between pseudo-steady-state transfer and transient transfer models. The proposed model can simulate reservoirs considering stochastic permeability distribution.

58 GEOSCIENCES↗

Critical analysis of velocimetry methods for particulate flows from synthetic data

Particle tracking methods that extract high-fidelity particle velocity data from high speed video of particle laden flows is a common experimental technique applied to chemical processes. These measurements are used to better understand the motion of particles and fluids in complex systems and create data against which computational models are validated. However, the methods, codes, and experimental setups all have limitations. It is imperative that practitioners verify the methods and their implementation as well as understand the limitations of experimental setups. This work focuses on quantifying the visible depth of field in a high particle concentration fluidized bed. Following a precedent set by the particle imaging velocimetry community, a particle velocity field is manufactured using a computational fluid dynamics and discrete element method simulation. Photo realistic high-speed videos are rendered based on the simulated data using the three-dimensional creation software Blender. Particle velocities are extracted from the synthetic high-speed videos using three variants of Particle Tracking Velocimetry and Optical Flow Velocimetry methodologies. Here, the tracked results are then compared to the known solution, quantifying the error associated with the assumed visible depth. The results indicate that at depth of one particle diameter, all three particle tracking codes give accurate measurements, largely within 5%. However, the error increases when the full bed video measurements are compared to the known solution at one particle diameter, i.e., mimicking a validation study. Finally, for some statistics the constant depth assumption only increases the error slightly, for others significantly.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A spectral Galerkin method for the coupled Orr-Sommerfeld and induction equations for free-surface MHD.

We develop and test spectral Galerkin schemes to solve the coupled Orr-Sommerfeld and induction equations for parallel, incompressible MHD in free-surface and fixed-boundary geometries. The schemes discrete bases consist of Legendre internal shape functions, supplemented with nodal shape functions for the weak imposition of the stress and insulating boundary conditions. The orthogonality properties of the basis polynomials solve the matrix-coefficient growth problem, and eigenvalue-eigenfunction pairs can be computed stably at spectral orders at least as large as p=3000 with p-independent roundoff error. Accuracy is limited instead by roundoff sensitivity due to non-normality of the stability operators at large hydrodynamic and/or magnetic Reynolds numbers (Re,Rm {ge} 4 x 10{sup 4}). In problems with Hartmann velocity and magnetic-field profiles we employ suitable Gauss quadrature rules to evaluate the associated exponentially weighted sesquilinear forms without error. An alternative approach, which involves approximating the forms by means of Legendre-Gauss-Lobatto quadrature at the 2p?1 precision level, is found to yield equal eigenvalues within roundoff error. As a consistency check, we compare modal growth rates to energy growth rates in nonlinear simulations and record relative discrepancy smaller than 10{sup -5} for the least stable mode in free-surface flow at Re = 3 x 10{sup 4}. Moreover, we confirm that the computed normal modes satisfy an energy conservation law for free-surface MHD with error smaller than 10{sup -6}. The critical Reynolds number in free-surface MHD is found to be sensitive to the magnetic Prandtl number Pm, even at the Pm=O(10{sup -5}) regime of liquid metals.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Machine learning changes the rules for flux limiters

Learning to integrate non-linear equations from highly resolved direct numerical simulations has seen recent interest for reducing the computational load for fluid simulations. Here, we focus on determining a flux-limiter for shock capturing methods. Focusing on flux limiters provides a specific plug-and-play component for existing numerical methods. Since their introduction, an array of flux limiters has been designed. Using the coarse-grained Burgers' equation, we show that flux-limiters may be rank-ordered in terms of their log-error relative to high-resolution data. We then develop a theory to find an optimal flux-limiter and present flux-limiters that outperform others tested for integrating Burgers' equation on lattices with [Formula: see text], and 2x, 3x, 4x, and 8x coarse-grainings. We train a continuous piecewise linear limiter by minimizing the mean-squared misfit to six-grid point segments of high-resolution data, averaged over all segments. While flux limiters are generally designed to have an output of φ(r) = 1 at a flux ratio of r = 1, our limiters are not bound by this rule and yet produce a smaller error than standard limiters. Here we find that our machine learned limiters have distinctive features that may provide new rules-of-thumb for the development of improved limiters. Additionally, we use our theory to learn flux-limiters that outperform standard limiters across a range of values (as opposed to at a specific fixed value) of coarse-graining, number of discretized bins, and diffusion parameter. This demonstrates the ability to produce flux limiters that should be more broadly useful than standard limiters for general applications.

97 MATHEMATICS AND COMPUTING↗

Lattice study of N=2 Landau-Ginzburg model using a Nicolai map

It has been conjectured that the two-dimensional N=2 Wess-Zumino model with a quasihomogeneous superpotential provides the Landau-Ginzburg description of the N=2 superconformal minimal models. For the cubic superpotential W={lambda}{Phi}{sup 3}/3, it is expected that the Wess-Zumino model describes A{sub 2} model and the chiral superfield {Phi} shows the conformal weight (h,h)=(1/6,1/6) at the IR fixed point. We study this conjecture by a lattice simulation, extracting the weight from the finite volume scaling of the susceptibility of the scalar component in {Phi}. We adopt a lattice model with the overlap fermion, which possesses a Nicolai map and a discrete R-symmetry. We set a{lambda}=0.3 and generate the scalar field configurations by solving the Nicolai map on LxL lattices in the range L=18-32. To solve the map, we use the Newton-Raphson algorithm with various initial configurations. The result is 1-h-h=0.660{+-}0.011, which is consistent with the conjecture within the statistical error, while a systematic error is estimated as less than 0.5%.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Full field imaging based instantaneous hyperspectral absolute refractive index measurement

Multispectral refractometers typically measure refractive index (RI) at discrete monochromatic wavelengths via a serial process. We report on the demonstration of a white light full field imaging based refractometer capable of instantaneous multispectral measurement of absolute RI of clear liquid/gel samples across the entire visible light spectrum. The broad optical bandwidth refractometer is capable of hyperspectral measurement of RI in the range 1.30 1.70 between 400nm 700nm with a maximum error of 0.0036 units (0.24% of actual) at 414nm for a = 1.50 sample. We present system design and calibration method details as well as results from a system validation sample.

99 GENERAL AND MISCELLANEOUS↗

Mapping snow depth and volume at the alpine watershed scale from aerial imagery using Structure from Motion

Time series mapping of snow volume in the mountains at global scales and at resolutions needed for water resource management is an unsolved challenge to date. Snow depth mapping by differencing surface elevations from airborne lidar is a mature measurement approach filling the observation gap operationally in a few regions, primarily in mountain headwaters in the Western United States. The same concept for snow depth retrieval from stereo- or multi-view photogrammetry has been demonstrated, but these previous studies had limited ability to determine the uncertainties of photogrammetric snow depth at the basin scale. For example, assessments used non-coincident or discrete points for reference, masked out vegetation, or compared a subset of the fully snow-covered study domain. Here, using a unique data set with simultaneously collected airborne data, we compare snow depth mapped from multi-view Structure from Motion photogrammetry to that mapped by lidar at multiple resolutions over an entire mountain basin (300 km 2 ). After excluding reconstruction errors (negative depths), SfM had lower snow-covered area (~27%) and snow volume (~16%) compared to lidar. The reconstruction errors were primarily in areas with vegetation, shallow snow (< 1 m), and steep slopes (> 60°C). Across the overlapping snow extent, snow depths compared well to lidar with similar mean values (< 0.03 m difference) and snow volume (± 5%) for output resolutions of 3 m and 50 m, and with a normalized median absolute deviation of 0.19 m. Our results indicate that photogrammetry from aerial images can be applied in the mountains but would perform best for deeper snowpacks above tree line.

54 ENVIRONMENTAL SCIENCES↗

Methods for diffusive relaxation in the Pn equation

We present recent progress in the development of two substantially different approaches for simulating the so-called of P{sub N} equations. These are linear hyperbolic systems of PDEs that are used to model particle transport in a material medium, that in highly collisional regimes, are accurately approximated by a simple diffusion equation. This limit is based on a balance between function values and gradients of certain variables in the P{sub N} system. Conventional reconstruction methods based on upwinding approximate such gradients with an error that is dependent on the size of the computational mesh. Thus in order to capture the diffusion limit, a given mesh must resolve the dynamics of the continuum equation at the level of the mean-free-path, which tends to zero in the diffusion limit. The two methods analyzed here produce accurate solutions in both collisional and non-collisional regimes; in particular, they do not require resolution of the mean-free-path in order to properly capture the diffusion limit. The first method is a straight-forward application of the discrete Galerkin (DG) methodology, which uses additional variables in each computational cell to capture the balance between function values and gradients, which are computed locally. The second method uses a temporal splitting of the fast and slow dynamics in the P{sub N} system to derive so-called regularized equations for which the diffusion limit is built-in. We focus specifically on the P{sub N} equations for one-dimensional, slab geometries. Preliminary results for several benchmark problems are presented which highlight the advantages and disadvantages of each method. Further improvements and extensions are also discussed.

97 MATHEMATICS AND COMPUTING↗

Modeling the effect of wind speed and direction shear on utility‐scale wind turbine power production

Abstract Wind speed and direction variations across the rotor affect power production. As utility‐scale turbines extend higher into the atmospheric boundary layer (ABL) with larger rotor diameters and hub heights, they increasingly encounter more complex wind speed and direction variations. We assess three models for power production that account for wind speed and direction shear. Two are based on actuator disc representations, and the third is a blade element representation. We also evaluate the predictions from a standard power curve model that has no knowledge of wind shear. The predictions from each model, driven by wind profile measurements from a profiling LiDAR, are compared to concurrent power measurements from an adjacent utility‐scale wind turbine. In the field measurements of the utility‐scale turbine, discrete combinations of speed and direction shear induce changes in power production of −19% to +34% relative to the turbine power curve for a given hub height wind speed. Positive speed shear generally corresponds to over‐performance and increasing magnitudes of direction shear to greater under‐performance, relative to the power curve. Overall, the blade element model produces both higher correlation and lower error relative to the other models, but its quantitative accuracy depends on induction and controller sub‐models. To further assess the influence of complex, non‐monotonic wind profiles, we also drive the models with best‐fit power law wind speed profiles and linear wind direction profiles. These idealized inputs produce qualitative and quantitative differences in power predictions from each model, demonstrating that time‐varying, non‐monotonic wind shear affects wind power production.

Energy & Fuels↗

High-order shock-fitted detonation propagation in high explosives

In this paper, a highly accurate numerical shock and material interface fitting scheme composed of fifth-order spatial and third- or fifth-order temporal discretizations is applied to the two-dimensional reactive Euler equations in both slab and axisymmetric geometries. High rates of convergence are not typically possible with shock-capturing methods as the Taylor series analysis breaks down in the vicinity of discontinuities. Furthermore, for typical high explosive (HE) simulations, the effects of material interfaces at the charge boundary can also cause significant computational errors. Fitting a computational boundary to both the shock front and material interface (i.e. streamline) alleviates the computational errors associated with captured shocks and thus opens up the possibility of high rates of convergence for multi-dimensional shock and detonation flows. Several verification tests, including a Sedov blast wave, a Zel'dovich–von Neumann–Döring (ZND) detonation wave, and Taylor–Maccoll supersonic flow over a cone, are utilized to demonstrate high rates of convergence to nontrivial shock and reaction flows. Comparisons to previously published shock-capturing multi-dimensional detonations in a polytropic fluid with a constant adiabatic exponent (PF-CAE) are made, demonstrating significantly lower computational error for the present shock and material interface fitting method. For an error on the order of 10 m / s , which is similar to that observed in experiments, shock-fitting offers a computational savings on the order of 1000. In addition, the behavior of the detonation phase speed is examined for several slab widths to evaluate the detonation performance of PBX 9501 while utilizing the Wescott–Stewart–Davis (WSD) model, which is commonly used in HE modeling. It is found that the thickness effect curve resulting from this equation of state and reaction model using published values is dramatically more steep than observed in recent experiments. Utilizing the present fitting strategy, in conjunction with a nonlinear optimizer, a new set of reaction rate parameters improves the correlation of the model to experimental results. In conclusion, this new model is tested against two dimensional slabs as a validation test.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗

Neutrino Flavor Model Building and the Origins of Flavor and CP Violation

The neutrino sector offers one of the most sensitive probes of new physics beyond the Standard Model of Particle Physics (SM). The mechanism of neutrino mass generation is still unknown. The observed suppression of neutrino masses hints at a large scale, conceivably of the order of the scale of a rand unified theory (GUT), which is a unique feature of neutrinos that is not shared by the charged fermions. The origin of neutrino masses and mixing is part of the outstanding puzzle of fermion masses and mixings, which is not explained ab initio in the SM. Flavor model building for both quark and lepton sectors is important in order to gain a better understanding of the origin of the structure of mass hierarchy and flavor mixing, which constitute the dominant fraction of the SM parameters. Recent activities in neutrino flavor model building based on non-Abelian discrete flavor symmetries and modular flavor symmetries have been shown to be a promising direction to explore. The emerging models provide a framework that has a significantly reduced number of undetermined parameters in the flavor sector. In addition, such a framework affords a novel origin of CP violation from group theory due to the intimate connection between physical CP transformation and group theoretical properties of non-Abelian discrete groups. Model building based on non-Abelian discrete flavor symmetries and their modular variants enables the particle physics community to interpret the current and anticipated upcoming data from neutrino experiments. Non-Abelian discrete flavor symmetries and their modular variants can result from compactification of a higher-dimensional theory. Pursuit of flavor model building based on such frameworks thus also provides the connection to possible UV completions: in particular, to string theory. We emphasize the importance of constructing models in which the uncertainties of theoretical predictions are smaller than, or at most compatible with, the error bars of measurements in neutrino experiments. While there exist proof-of-principle versions of bottom-up models in which the theoretical uncertainties are under control, it is remarkable that the key ingredients of such constructions were discovered first in top-down model building. We outline how a successful unification of bottom-up and top-down ideas and techniques may guide us towards a new era of precision flavor model building in which future experimental results can give us crucial insights into the UV completion of the SM.

Almumin, Yahya (ORCID:0000000196396772)↗

MPAS-Seaice (v1.0.0): sea-ice dynamics on unstructured Voronoi meshes

Abstract. We present MPAS-Seaice, a sea-ice model which uses the Model for Prediction Across Scales (MPAS) framework and spherical centroidal Voronoi tessellation (SCVT) unstructured meshes. As well as SCVT meshes, MPAS-Seaice can run on the traditional quadrilateral grids used by sea-ice models such as CICE. The MPAS-Seaice velocity solver uses the elastic–viscous–plastic (EVP) rheology and the variational discretization of the internal stress divergence operator used by CICE, but adapted for the polygonal cells of MPAS meshes, or alternatively an integral (“finite-volume”) formulation of the stress divergence operator. An incremental remapping advection scheme is used for mass and tracer transport. We validate these formulations with idealized test cases, both planar and on the sphere. The variational scheme displays lower errors than the finite-volume formulation for the strain rate operator but higher errors for the stress divergence operator. The variational stress divergence operator displays increased errors around the pentagonal cells of a quasi-uniform mesh, which is ameliorated with an alternate formulation for the operator. MPAS-Seaice shares the sophisticated column physics and biogeochemistry of CICE and when used with quadrilateral meshes can reproduce the results of CICE. We have used global simulations with realistic forcing to validate MPAS-Seaice against similar simulations with CICE and against observations. We find very similar results compared to CICE, with differences explained by minor differences in implementation such as with interpolation between the primary and dual meshes at coastlines. We have assessed the computational performance of the model, which, because it is unstructured, runs with 70 % of the throughput of CICE for a comparison quadrilateral simulation. The SCVT meshes used by MPAS-Seaice allow removal of equatorial model cells and flexibility in domain decomposition, improving model performance. MPAS-Seaice is the current sea-ice component of the Energy Exascale Earth System Model (E3SM).

58 GEOSCIENCES↗

Robust system size reduction of discrete fracture networks: a multi-fidelity method that preserves transport characteristics

In this paper, we propose a multi-fidelity system reduction technique that uses weighted graphs paired with three-dimensional discrete fracture network (DFN) modelling for efficient simulation of subsurface flow and transport in fractured media. DFN models are used to simulate flow and transport in subsurface fractured rock with low-permeability. One method to alleviate the heavy computational overhead associated with these simulations is to reduce the size of the DFN using a graph representation of it to identify the primary flow sub-network and only simulate flow and transport thereon. The first of these methods used unweighted graphs constructed solely on DFN topology and could be used for accurate predictions of first-passage times. However, these techniques perform poorly when predicting later stages of the mass breakthrough. We utilize a weighted-graph representation of the DFN where edge weights are based on hydrological parameters in the DFN that allows us to exploit the kinematic quantities derivable a posteriori from the flow solution obtained on the graph representation of the DFN to perform system reduction and predict the later stages of the breakthrough curve with high fidelity. We also propose and demonstrate the use of an adaptive pruning algorithm with error control that produces a pruned DFN sub-network whose predicted mass breakthrough agrees with the original DFN within a user-specified tolerance. Lastly, the method allows for the level of accuracy to be a user-controlled parameter.

58 GEOSCIENCES↗

Short‐term time step convergence in a climate model

Abstract This paper evaluates the numerical convergence of very short (1 h) simulations carried out with a spectral‐element (SE) configuration of the Community Atmosphere Model version 5 (CAM5). While the horizontal grid spacing is fixed at approximately 110 km, the process‐coupling time step is varied between 1800 and 1 s to reveal the convergence rate with respect to the temporal resolution. Special attention is paid to the behavior of the parameterized subgrid‐scale physics. First, a dynamical core test with reduced dynamics time steps is presented. The results demonstrate that the experimental setup is able to correctly assess the convergence rate of the discrete solutions to the adiabatic equations of atmospheric motion. Second, results from full‐physics CAM5 simulations with reduced physics and dynamics time steps are discussed. It is shown that the convergence rate is 0.4—considerably slower than the expected rate of 1.0. Sensitivity experiments indicate that, among the various subgrid‐scale physical parameterizations, the stratiform cloud schemes are associated with the largest time‐stepping errors, and are the primary cause of slow time step convergence. While the details of our findings are model specific, the general test procedure is applicable to any atmospheric general circulation model. The need for more accurate numerical treatments of physical parameterizations, especially the representation of stratiform clouds, is likely common in many models. The suggested test technique can help quantify the time‐stepping errors and identify the related model sensitivities.

54 ENVIRONMENTAL SCIENCES↗

Multi-resolution partial differential equations preserved learning framework for spatiotemporal dynamics

Traditional data-driven deep learning models often struggle with high training costs, error accumulation, and poor generalizability in complex physical processes. Physics-informed deep learning (PiDL) addresses these challenges by incorporating physical principles into the model. Most PiDL approaches regularize training by embedding governing equations into the loss function, yet this depends heavily on extensive hyperparameter tuning to weigh each loss term. To this end, we propose to leverage physics prior knowledge by “baking” the discretized governing equations into the neural network architecture via the connection between the partial differential equations (PDE) operators and network structures, resulting in a PDE-preserved neural network (PPNN). This method, embedding discretized PDEs through convolutional residual networks in a multi-resolution setting, largely improves the generalizability and long-term prediction accuracy, outperforming conventional black-box models. The effectiveness and merit of the proposed methods have been demonstrated across various spatiotemporal dynamical systems governed by spatiotemporal PDEs, including reaction-diffusion, Burgers’, and Navier-Stokes equations.

97 MATHEMATICS AND COMPUTING↗