Simulation Science Lab: Computational Materials Science

German Research School for Simulation Sciences and RWTH Aachen University
WS 2013/14, Fr 09:45-12:00, Lecture Hall, GRS Jülich
note change in place and time!
CAMPUS link


Lectures

  1. Introduction
    preparation:
    1. clone SiScLab13 to your laptop and RWTH cluster account
    2. read Sec. 1 and 2.1 on the Lanczos method
    3. if you need a refresher on second quantization, read Sec. 1, 2, and 4 on many-body states
    Lanczos method
    tight-binding approximation
    many-electron states
    Hubbard model
  2. Many-body hopping
    setting up lookup tables for basis states
    counting the number of set bits
    operations on the many-body states: bit-operations
    hopping Hamiltonian on the many-body Hilbert space
    sparse matrix representation
  3. Many-body states
    disassembling your code and checking intrinsics
    sparse representation of many-body hopping
    Slater determinants
    representation of many-body states
  4. Lanczos iteration
    allocating large vectors
    basic linear algebra operations
    compiler optimizations
    Lanczos iteration
  5. Debugging
    example codes: basic.c and test.c
  6. Cache
    interpreting profiles: number of calls, timing/scaling
    Lieb-Wu solution for the infinite Hubbard-chain
    cache optimization
  7. Cache in Practice:
    cachegrind
  8. Trading memory access for computation
    Hubbard interaction without lookup tables

Erik Koch