Research News

Fall 2026 — NVIDIA Academic Grant Award

Our group has been awarded a hardware grant from the NVIDIA Academic Grant Program, receiving two state-of-the-art NVIDIA RTX PRO 6000 GPUs to support our ongoing research on efficient, GPU-accelerated numerical methods for large-scale phase-field simulations. This award reflects a sustained research trajectory in GPU-based scientific computing, highlighted by the following papers:

• Orizaga, S., Fabien, M., Millard, M. Efficient numerical approaches with accelerated graphics processing unit (GPU) computations for Poisson problems and Cahn-Hilliard equations. AIMS Mathematics, 2024.
  Introduced GPU-accelerated solvers for Poisson and Cahn-Hilliard problems, establishing the foundation for our group's transition to consumer-GPU-based scientific computing. https://www.aimspress.com/article/doi/10.3934/math.20241334

• Orizaga, S. An energy-stable spectral method for the block copolymer equation via the biharmonic modified approach. Computational Materials Science, 2026.
  A sole-authored study introducing the Biharmonic Modified (BHM) scheme for the Block Copolymer equation, achieving a 40× CPU-to-GPU speedup on consumer laptop hardware for large-scale 3D simulations at 256³ resolution, with rigorous energy stability guarantees and machine-precision mass conservation. https://www.sciencedirect.com/science/article/pii/S0927025626004726

• Orizaga, S., Yin, P., Choudhuri, D. A Convex Splitting Spectral Method for the Phase Field Crystal Equation: Energy Stability, Computational Stability Maps, and Three-Dimensional GPU Simulations. (Under review)
  Develops an unconditionally energy-stable scheme for the Phase Field Crystal equation, supported by a computational stability map derived from 40,000 GPU-accelerated simulations, and demonstrates 3D simulations at resolutions up to 512³ on a single consumer GPU. https://arxiv.org/abs/2607.25177

This award accelerates a dual-track research program at the forefront of GPU-accelerated computational mathematics, driving parallel breakthroughs in numerical theory and high-performance computing accessibility:

• The Consumer Hardware Frontier: We are breaking resolution barriers previously thought to require institutional HPC infrastructure, pioneering second-order, structurally compact, energy-stable spectral methods that bring elite three-dimensional computing directly to consumer-grade GPUs through open, intuitive frameworks.

• The Professional Multi-Scale Frontier: Leveraging this massive professional hardware infrastructure, our group is executing unprecedented, high-density parameter sweeps to systematically chart the true numerical stability landscapes of leading phase-field paradigms — including CS, SAV, and our own BHM framework. Through deep-time simulations, we are resolving the long-term thermodynamic and physical behavior of highly complex, multi-component systems that are mathematically inaccessible via closed-form analysis.

These developments represent our immediate architectural plans to redefine the limits of physical fidelity and democratize large-scale scientific computing worldwide.

.