Physics-constrained Gaussian Processes for Predicting Shockwave Hugoniot Curves
George D. Pasparakis, Himanshu Sharma, Rushik Desai, Chunyu Li, Alejandro Strachan, Lori Graham-Brady, Michael D. Shields · 2026
Read Physics-constrained Gaussian Processes for Predicting Shockwave Hugoniot Curves by George D. Pasparakis, Himanshu Sharma, Rushik Desai, Chunyu Li, Alejandro Strachan, Lori Graham-Brady, Michael D. Shields (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-06655.txt.
Introduction
George Pasparakis and six co-authors guess shock states of stuff with far less data in 2026. A shock curve lists the states a stuff reaches when hit by a shock. Each point on that curve has cost a pricey run on a large machine.
Their frame is a chance-based curve fit bound by the laws of shocks. It learns from just a few runs and fills in the rest of the curve. A chance form of a Taylor sum builds the fit at its core. Jump rules at the shock front shape the fit so it stays true to the laws. The fit thus honors heat laws, not just the dots on a graph. Labs that study blasts and impacts can stretch small budgets much more. The way is built for work where each run costs a great deal.
Catalog record
| CREATOR | George D. Pasparakis, Himanshu Sharma, Rushik Desai, Chunyu Li, Alejandro Strachan, Lori Graham-Brady, Michael D. Shields |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-06655
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Physics-constrained Gaussian Processes for Predicting Shockwave Hugoniot Curves George D. Pasparakis, Himanshu Sharma, Rushik Desai, Chunyu Li, Alejandro Strachan, Lori Graham-Brady, Michael D. Shields Abstract A physics-constrained Gaussian Process regression framework is developed for predicting shocked material states along the Hugoniot curve using data from a small number of shockwave simulations. The proposed Gaussian process employs a probabilistic Taylor series expansion in conjunction with the Rankine–Hugoniot jump conditions between the various shocked material states to construct a thermodynamically consistent covariance function. This leads to the formulation of an optimization problem over a small number of interpretable hyperparameters and enables the identification of regime transitions, from a leading elastic wave to trailing plastic and phase transformation waves. This work is motivated by the need to investigate shock-driven material response for materials discovery and for offering mechanistic insights in regimes where experimental characterizations and simulations are costly. The proposed methodology relies on large-scale molecular dynamics which are an accurate but expensive computational alternative to experiments. Under these constraints, the proposed methodology establishes Hugoniot curves from a limited number of molecular dynamics simulations. We consider silicon carbide as a representative material and atomic-level simulations are performed using a reverse ballistic approach together with appropriate interatomic potentials. The framework reproduces the Hugoniot curve with satisfactory accuracy while also quantifying the uncertainty in the predictions using the Gaussian Process posterior. 1 Introduction Large-amplitude wave propagation in solids induced by a high-velocity impact is governed by the constitutive behavior of materials. The response of the material to the dynamical insult encompasses a broad range of physical phenomena, including elasticity, thermodynamics, phase transformations, plasticity, and distinct failure mechanisms. Experimental investigation and analysis of shockwave propagation provide a powerful means of understanding this material response under extreme dynamic conditions and can offer critical insights into the underlying physics. However, experimental characterization of shock response poses numerous challenges. A considerable level of complexity arises due to the sensitivity in loading parameters such as sample confinement, impact angle, sample alignment, and crystal orientation relative to the shock propagation, resulting in a large number of experimental scenarios [ 30 ] . This is coupled with extremely fast time scales, in which the material transitions from the unloaded to the shocked state in a matter of picoseconds. This transition manifests itself as a discontinuity of near-atomic thickness between the two states, which is very difficult to measure accurately. Further complicating the process, the resulting extreme pressures and temperatures result in plastic deformation, phase transitions, and chemical reactions. To complement experiments, the research community developed advanced computational models. Molecular dynamics (MD) is a particularly powerful method that can provide insights into the underlying physics of the shock response [ 13 ] . Early work on metals revealed the atomistic mechanisms behind plastic deformation [ 14 ] , phase transitions [ 17 ] and dynamic failure [ 24 ] . Recently, larger-scale simulations and more broadly applicable interatomic potentials have enabled the simulation of more complex materials and physico-chemical processes [ 27 , 9 , 32 ] . Despite demonstrating good agreement with experimental data, see e.g. [ 16 ] , large-scale MD simulations can be computationally demanding. Shock simulations typically scale quadratically with system length as increasing the length of the simulation requires increasing the simulation time. A typical 200 M atom shock simulation requires \sim 7000 CPU hours for 40000 steps with 1fs timestep using AMD Epyc 7763 “Milan” CPUs @ 2.2GHz. To mitigate the computational burden, machine learning methods can be employed to generate shock response data and, more broadly, to predict material behavior. For example, neural network-based approaches have shown promise in developing interatomic potentials [ 2 ] , generating samples from equilibrium distributions [ 19 ] , predicting full-field material responses [ 20 ] , and mapping microstructure to thermal response under shock loading [ 18 ] . However, neural network-based approaches require a large amount of training data, which is often difficult or infeasible to obtain. Furthermore, these approaches are not accompanied by uncertainty quantification (UQ) capabilities. Accounting for this limitation, in this study a Gaussian process regression (GPR) framework is developed as a data-efficient, non-parametric Bayesian alternative method that not only performs well with limited training samples but also naturally provides predictive uncertainty estimates [ 31 ] . The reliance on training data can be further reduced by incorporating physical constraints into the GP framework, which also ensures realistic predictions. Recent studies [ 12 , 22 ] have introduced physics-constrained GPR frameworks for data-driven materials modeling and demonstrated their effectiveness under limited data conditions. Motivated by these developments, a physics-constrained GPR framework is proposed for predicting shocked material states along the Hugoniot. This approach applies to a broad class of materials by adhering to fundamental conservation laws across the shock front. Specifically, the framework enforces the Rankine–Hugoniot conditions as physical constraints by deriving a physics-based covariance function, constructed using a Taylor series expansion, that inherently satisfies these conditions. The proposed method is capable of modeling the transition from purely elastic shock response to elastic response with a trailing plastic wave, and ultimately into the overdriven regime. The model relies on a small number of interpretable parameters and is analytically shown to be consistent with the laws of thermodynamics. The model is applied to Silicon Carbide (SiC) as a representative material. SiC is a highly versatile engineering material that possesses exceptional physical properties, such as high thermal conductivity, stiffness, and hardness along with low density. It is shown that the proposed method can model SiC Hugoniot behavior in a computationally inexpensive manner using limited MD simulation data while also quantifying the uncertainty in the estimated state. 2 Background Shock-compression occurs when a material is subjected to an abrupt, high-speed impact or loading event, whose duration is considerably shorter than the characteristic time for the material to respond through inertial motion. For sufficiently high impact velocities, a structured wave with very short rise time (sub-picoseconds within single crystals) forms such that the thickness of the wave front can be ignored or cannot be measured experimentally. This leads to a jump in the principal mechanical and thermodynamic variables such as pressure, strain, velocity, and temperature. A schematic representation of shock wave loading is shown in Fig. 1 , where a projectile impacts the material at piston velocity u_{p} . The shock wave travels through the medium with velocity u_{s} , which is hypersonic relative to the unperturbed material, while the material behind the shock front moves at the particle velocity \nu_{z} relative to a stationary frame of reference (shown here as the initial boundary of the undeformed material). Figure 1 : Schematic representation of shock wave propagation in a material element. In a shock wave, gradients of the thermomechanical parameters in the unperturbed regions of the material are negligible compared to gradients at the shock front. This gives rise to a jump at the shock front. For steady-state shocks, applying conservation of mass, momentum, and energy across the jump gives relationships between the material state ahead of the shock (the “initial state”) and the state behind the shock front (the “final state”). These are commonly referred to as the Rankine-Hugoniot jump conditions, in which the shock wave is characterized by