2002 Fiscal Year Final Research Report Summary
Development of Lattice QCD Simulation System on PC Cluster
| Project/Area Number |
12554008
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 展開研究 |
|---|
| Research Field |
素粒子・核・宇宙線
|
|---|
| Research Institution | HIROSHIMA UNIVERSITY |
|---|
Principal Investigator |
NAKAMURA Atsushi Hiroshima University, Information Media Center, Professor, 情報メディア教育研究センター, 教授 (30130876)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
IKESAKA Morio Fujitsu Laboratory, Senior Researcher, HPC研究センター, 主管研究員
HIOKI shinji Tedukayama university, Faculty of Economy and Information Professor, 経営情報学部, 教授 (70238252)
SHOJI Fumiyoshi Hiroshima University, Information Media Center, Assistant, 情報メディア教育研究センター, 助手 (60310504)
|
|---|
| Project Period (FY) |
2000 – 2002
|
| Keywords | QCD / Fortarn90 / parallel code / PC cluster |
|---|
| Research Abstract |
QCD (Quantum Chromo-Dynamics) is the theory of quarks and gluons. Numerical simulations of lattice QCD formulated by Wilson are the most powerful and reliable method to study QCD. Due to the fast development of computers, I.e., the price becomes lower and the speed increases rapidly, the lattice QCO simulations on PC cluster become possible, many researchers in elementary/nuclear physics are interested in the method.
Although many theoretical physicists in elementary particle and nuclear physics understand lattice QCD well, it is difficult for them to perform lattice QCD simulations, since it required many knowledge of numerical simulation techniques and long time. In this project wehlave developed an open source code of lattice QCD in Fortran 90. The code is called Lattice Tool Kit in Fortran90 (LTK f90) and can be found at
http://nio-mon.riise.hiroshirna-u.ac.jp/~LTK/
Thanks to the module in Fortran90, we can write programs which are similar to the algorithm and easy to understand. Because Fortran90 sometime depends on the compiler of the machine, we have prepared a user-manual, which can also be obtained on the above web.
We use MPI (Message Passing Interface) to parallelize the code,and then large lattice size simulations are possible.
MPI is the most effective way for parallel programming, but it is not so easy for non-specialist, and also a code with MPI can not be executed under non-MPI compiler. We are now developing a HPF (High Performance Fortran) version of LTK.
|
