CESMIX brings together researchers in numerical algorithms and scientific computing, quantum chemistry, materials science, and computer science to connect quantum and molecular simulations of materials with advanced programming languages, compiler technologies, and software performance engineering tools, underpinned by rigorous approaches to statistical inference and uncertainty quantification.

We are advancing the ability to predict the degradation (oxidation and melting) of complex (disordered and multi-component) materials under extreme loading, inaccessible to direct experimental observation. This application represents a technology domain of intense current interest, and has clear extensions to a much wider class of problems—i.e., those involving material interfaces in extreme environments.

Indeed, interfaces are crucial to our ability to understand, engineer, and control material behavior. Yet the predictive simulation of interface properties and their impact remains extremely challenging. Capturing the impact of interfaces requires resolving interfacial atomic structure and thus appealing to first principles (i.e., quantum mechanical) simulations; at the same time, interfacial structures are affected by conditions (e.g., thermal, mechanical, and chemical loading) at the engineering scale. This problem thus presents an enormous and fundamental multiscale simulation challenge, with myriad applications—to hypersonics and far beyond.

As an exemplar of our overaching goal, we consider hypersonics; our aim is to simulate materials exposed to ultra-high temperatures, extreme heat fluxes, and oxidative environments, as on the leading edges of hypersonic vehicles. Significant research has been devoted to the development of better materials for this application.

Diborides such as HfB2 and ZrB2 are attractive due to their high melting temperatures, but their oxidation resistance is not ideal. High-entropy alloys of diborides promise to combine both temperature and oxidation resistance with mechanical strength. But predicting these material properties is enormously difficult; indeed, high-entropy ceramics and glassy materials with complex interfaces and oxidative processes present major challenges to available methodology.

We are addressing these challenges by developing a comprehensive new multiscale materials simulation framework, bridging from multiple levels of quantum mechanical theory to hybrid quantum/classical methods to classical molecular dynamics.

Our target problem drives the development of new exascale computing capabilities in the broadest sense: compiler technologies aimed at both portability and composable performance on heterogeneous architectures; high-level and domain-specific languages; differentiable programming with new and legacy software; and new toolchains that facilitate uncertainty quantification and inference across scales. Taking advantage of these capabilities, we are developing ab initio computational methods for modeling amorphous materials at high temperatures, superior inference-based methods for coarse-graining quantum mechanical interaction potentials with quantified uncertainty, and more accurate and efficient hypersonic flow computations. Our simulations are supported by a unique in-house source of high-temperature validation data.

To generate realistic loading conditions for our materials, we consider the aerothermal environment produced by a hypersonic shock-shock interaction around a circular cylinder. This hypersonic flow is representative of conditions at scramjet leading edges; more importantly, it is highly sensitive to the configuration of the shock structures and can induce extreme and localized heating on the material. In these situations, material damage starts at the surface and propagates into the bulk. Damage interfaces are then crucial to our ability to understand, engineer, and control material behavior. Predictive simulation of these interfaces—their properties, evolution, and impact—is thus our core challenge.

Meeting this challenge requires advances in modeling and simulation, uncertainty quantification (UQ) and verification/validation, exascale computing, and integration science, supported by an end-to-end software effort. Our predictive simulation and UQ frameworks are closely integrated, with UQ supporting a “decision engine” that determines how to optimally deploy and combine models to reach a targeted uncertainty in a system-level prediction; this simulation/UQ framework is illustrated in the figure at left. These approaches are realized on exascale hardware via multiple coordinated computer science innovations: semantic augmentation to enable differentiable programming with legacy codebases, high-level and domain-specific languages, and compiler technologies aimed at both portability and composable performance on heterogeneous architectures.