Mohammed Sayyari

Mohammed Sayyari

Postdoctoral Research Associate

Department of Mathematics and Statistics Old Dominion University

Biography

My name is Mohammed Sayyari; I work on positivity-preserving provably stable high-fidelity numerical schemes for the Navier-Stokes equations as a postdoc in the Department of Mathematics and Statistics at Old Dominion University. With a strong background in computer science, my current project integrates efficient and scalable computation with rigorous mathematical analysis to preserve the stable structure of the underlying model. Through international collaborations, I have contributed to impactful research, such as in the relaxation Runge-Kutta method, where we extended the method to include nonlinear convex functionals such as mathematical entropy. This method provides provable stability for fully-discrete schemes, in contrast to the stability of only the discrete spatial operators, common in this area of research. My long-term research plan is to develop provably stable, robust, and efficient schemes on high-performance computing (HPC) platforms for problems in fire simulation and numerical weather prediction (NWP). These schemes enhance the accuracy and stability of simulations for critical applications like climate modeling and aerospace engineering.

In addition to my research, I have consistently been passionate about teaching. Even when it was not required, I served as a Teaching Assistant during my PhD for a variety of courses spanning computer science and mathematics. For example, I assisted in creating homework and held office hours for courses such as Numerical Analysis of PDEs and Numerical Optimization. Additionally, I inaugurated the course of Numerical Methods for Internal Aerodynamics at Ruhr-Universität Bochum, creating all the course content, from slides and notes to a unique problem set and computer programming examples. I am committed to continuing learning and researching pedagogical methods. For example, I attended a course on the eight practices in teaching mathematics, where I learned key practices such as productive struggle, purposeful questioning, and eliciting evidence of student thinking.

Interests

  • Numerical Analysis of PDEs
  • Numerical Linear Algebra
  • Numerical Optimization
  • Evolutionary PDEs
  • Mathematics for Sustainability

Education

  • PhD in Applied Mathematics, 2022

    King Abdullah Universty of Science and Technology

  • MSc in Applied Mathematics, 2018

    King Abdullah Universty of Science and Technology

  • BSc in Computer Science with a minor in Mathematics, 2016

    Kansas State University

Experience

 
 
 
 
 

Postdoc Research Associate

Department of Mathematics and Statistics

Feb 2024 – Jan 2025 Virginia

Role includes:

  • Positivity Preserving Numerical Schemes
 
 
 
 
 

Postdoc Research Assistant

Lehrstuhl für Thermische Turbomaschinen und Flugtriebwerke, Ruhr-Universität Bochum

Aug 2022 – Dec 2023 Germany

Role includes:

  • Scalability of CFD solver libraries
  • Supervising students
  • Lecturing on CFD in practice

Recent Publications

(2024). Fully-Discrete Lyapunov Consistent Discretizations for Parabolic Reaction-Diffusion Equations with r Species. In CAMC.

Project Source Document DOI

(2024). Particle-resolved simulations and measurements of the flow through a uniform packed bed. In PoF.

Project Source Document DOI

(2023). Large-eddy simulation of a channel flow over an irregular porous matrix. In PAMM.

Project Source Document DOI

Awards

i-Center Undergraduate Scholar Award

Two well known problems in combinatorics are ”circle cut by chords” and ”circle cut by line. The first has two simple rules, each dot is a distance one away from any other dot on the circle, and any three edges cannot intersect at one point. It is a count of the number of regions as the number of dots increase. The second has one rule, any three edges, cuts in this context, cannot intersect at one point. It is a count of the number of regions as the number of cuts increase. Now, the interesting part is that we noticed that the second problem is part of a family of problems, Balls in any dimension n cut by hyperspaces in n −1. Then, we noted that in dimension n = 4, the number of regions per cut c matches, to some number of cuts, that of the number of regions per one less dot in the first problem. Our goal is to find a bijection between the two problems.
See certificate

Contact