Large Eigenvalue Problems
Federico Milano, Ioannis Dassios, Muyang Liu, Georgios Tzounas · 2020
The largest ever computer-based matrix eigenvalue calculations with dense eigensolvers to date were carried out in Japan in 2014. In particular, the analysis was carried out with the K computer in Riken, to evaluate the Japanese eigensolver EigenExa. EigenExa was compared with other parallel dense eigensolvers (such as ScaLAPACK). EigenExa managed to find the eigenvalues of a 10 6 × 10 6 https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9780429325311/eb3d7478-7cfc-41e6-94d2-fc1661ce66fa/content/math15_1.tif"/> problem in about 1 hour. This result has a high cost. The K computer includes 88,000 processors which draw a peak power of 12.6 MW (see Figure 15.1 ). 1 Its operation costs annually US$10 million.