High-performance computing for materials simulations
Hands-on Course given by Antimo Marrazzo
Term
2
Credit
3
Start Date
Room
131
Program
Schedule and teachers TBA.
Tentative program:
- Basic theory of high-performance computing
- Introduction to high-performance computing
- Parallel computing
- MPI and OpenMP paradigms
- GPU-based computing
- Strong/weak scaling, Amdahl’s and Gustafson’s laws,
- Throughput and latency parallelism
- FFT and linear algebra: scaling and parallel implementation
2. Parallel programming tutorial
- “For loop” example in MPI, OpenMP, openACC
- Parallel matrix multiplication, distributed Jacobi solver
3. Tutorial with parallel and GPU Quantum ESPRESSO
- Efficient and optimized usage of multiple levels of parallelism
- Band-structure and molecular-dynamics simulations of large and small systems on Tier-0 supercomputers
- GPU oversubscription with multi-process service (MPS) for Nvidia cards
4. Introduction to materials informatics
- Automation & databases
- High-throughput computing
- Brief overview of machine learning in computational materials science, with examples of applications
5. Tutorial with AiiDA-Quantum ESPRESSO
- Introduction to AiiDA
- Automated band-structure and phonon calculations with AiiDA-Quantum ESPRESSO
6. Hands-on projects with assistance and supervision