Abhinav Singh

Title:

A Context-Aware Language for Scalable Numerical Simulations of Topology Driven Morphogenesis.

Abstract:

Computer simulation of physical phenomena has led to multiple discoveries and has become a fundamental tool to study the behavior of complex biological systems. However, writing computer codes to perform simulations remains challenging and taxing. Implementing scalable and efficient codes on parallel computers of such simulations adds to the challenge. In this project, we aim to identify and implement abstractions for creating a domain-specific language that allows an effortless encoding of simulation programs to high-performance computers. Creating such a software system is difficult, as it needs to support various models and different numerical algorithms, all while maintaining scalability on parallel computer architectures. To aid this process, we developed a novel context-aware C++ expression system for scalable simulations using OpenFPM [1]. This generic expression system generates scalable code from mathematical equation models at compile time. Our new framework simplifies implementation and enables rapid testing of numerical codes without rewriting them significantly. It also cleanly separates the implementation of the model equations from that of the numerical algorithms. We plan to develop this high-performance numerical simulation framework further for simulating morphogenetic models in complex 3D geometries. Using these simulations, we aim to answer questions about the active fluid flow, such as the existence of active turbulence in 3D. The eventual goal of the project is to enable a comprehensive simulation framework that copes with biological shape complexity by using encoded roles of the language.

[1] A. Singh, P. Incardona, and I. F. Sbalzarini, “A C++ expression system for partial differential equations enables generic simulations of biological hydrodynamics,” Eur. Phys. J. E, vol. 44, no. 9, p. 117, Sep. 2021, doi: 10.1140/epje/s10189-021-00121-x.