Fast and accurate eigenvalue solver for many-core processors
| Project/Area Number |
21560065
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Engineering fundamentals
|
|---|
| Research Institution | Kobe University |
|---|
Principal Investigator |
YAMAMOTO Yusaku 神戸大学, 大学院・システム情報学研究科, 教授 (20362288)
|
|---|
| Project Period (FY) |
2009 – 2012
|
| Project Status |
Completed (Fiscal Year 2012)
|
| Budget Amount *help |
¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000)
Fiscal Year 2011: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2010: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2009: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
|
|---|
| Keywords | 固有値計算 / 線形計算 / メニーコア / 並列処理 / 固有値問題 / スーパーコンピュータ / ブロックヤコビ法 / 数値計算 / アルゴリズム / マルチコア / 誤差解析 / 直交化 / QR分解 |
|---|
| Research Abstract |
The objective of this research is to develop efficient algorithms for computing eigenvalues on many-core processors. The achievements of this research can be summarized as follows:
(1) Implementation and performance evaluation of an eigenvalue computation algorithm designed for multi-core/many-core processors
(2) Error analysis of the AllReduce algorithm for Householder QR decomposition
(3) A new highly parallel algorithm for incremental orthogonalization
(4) A new algorithm for computing the eigenvalues of a totally nonnegative band Hessenberg matrix with high relative accuracy
|
Report
(4 results)
Research Products
(39 results)
