| Abstract: |
Title: Course on Atomistic Modeling of Materials (MEMS 5612)
Primary Field of Science: Materials Research
Resource Requested: 100,000 SUs on SDSC Expanse Cluster
Abstract: This Education Allocation Request is to support a course MEMS 5612 “Atomistic Modeling of Materials” that the PI is currently teaching in the Spring 2025 semester at Washington University in St. Louis. The course is aimed primarily at graduate students and some senior undergraduates from the departments of Mechanical Engineering & Materials Science, Energy, Environmental & Chemical Engineering, Chemistry, and Physics & Astronomy at Washington University. It will introduce students to various computational methods for atomic-scale modeling of materials, including electronic structure calculations using density-functional theory (DFT), tight-binding, molecular dynamics (MD) simulations using classical force-fields, and Monte Carlo methods. It will also introduce them to Materials Informatics, including data mining from materials databases and using the scikit-learn package to build classification and regression models for select materials properties such as Youngs modulus. The course has three components: lectures, hands-on laboratories, and a final project. The basic background of these methods will be covered in the lectures along with examples of their use for calculating various material properties, such as mechanical, electronic, optical and magnetic properties. The laboratories will provide exposure to using powerful atomistic materials modeling codes, including DFT codes, such as Quantum Espresso, and MD codes, such as LAMMPS, to model the behavior of crystalline materials and polymers. Besides the lab component, students in groups of two will work on a final computational project that is expected to last 6 weeks.
This is the 8th time this course is being offered by the PI. Previously, students in the course have used the Comet and Expanse clusters at SDSC to work on projects, such as:
1. First-principles prediction of refractory high-entropy alloys
2. First-principles investigation of heavy-metal adsorption on 2D-MoS2
3. First-principles investigation of oxygen removal from CO2-rich gas stream, which was published as an research article in the journal Phys. Chem. Chem. Phys. (10.1039/D2CP04788H).
4. Molecular dynamics study of grain boundary-point defect interactions
5. Molecular-dynamics simulations of collisional growth of TiO2 nanoparticles
6. Prediction of glass-forming ability in metallic alloys with molecular dynamics
This course has been received well all the previous times it has been offered. In the past, several students have continued to expand on their final project for the course (with additional computational time from XSEDE, for instance, through the start-up allocation TG-CTS1900058 Mechanistic studies of autocatalytic oxygen removal from concentrated CO2-stream, PI: Pratim Biswas) as a side-project for their PhD studies. This has culminated in complete chapters in the PhD dissertation of students that are focused on atomic-scale modeling of relevant materials and processes, and manuscripts that are currently in review in peer-reviewed journals.
For Spring 2025, we have 8 students.
- There are 9 laboratory assignments in the course. Of these, 3 assignments that are focused on DFT calculations using Quantum Espresso require access to clusters. We expect 16,000 SUs (2000 SUs/student * 8 students) for the labs and assignments.
- For the final projects, there will be 3 groups (of 2-3 students/group). Based on previous experience, we expect ~28,000 SUs per group, leading to 84,000 SUs. We also note that both LAMMPS and Quantum Espresso are frequently used on the Expanse cluster and are well optimized. We note that both Quantum Espresso and LAMMPS are routinely used on Expanse and the codes are well optimized.
Therefore, we request a total of 100,000 SUs on the Expanse cluster at SDSC. The requested resources will benefit the students immensely to gain hands-on experience in modeling materials using state-of-the-art computing clusters.
|
0000-0003-1261-0087