dense eigensolvers?

Open discussion for MAGMA library (Matrix Algebra on GPU and Multicore Architectures)
Post Reply
huuanh2811
Posts: 1
Joined: Tue Jul 02, 2019 10:39 pm

dense eigensolvers?

Post by huuanh2811 » Tue Jul 02, 2019 10:40 pm

I found an SC15 poster that said all four dense symmetric tridiagonal eigensolvers (QR, divide and conquer, bisection and inverse iteration, and MRRR) were included in MAGMA. When I downloaded 2.0, however, I only saw the D&C driver. What is the actual status of the others?

mgates3
Posts: 892
Joined: Fri Jan 06, 2012 2:13 pm

Re: dense eigensolvers?

Post by mgates3 » Wed Jul 03, 2019 12:30 am

For complex matrices:
magma_[cz]heevd* uses divide-and-conquer
magma_[cz]heevr* uses MRRR
magma_[cz]heevx* uses QR iteration if MagmaRangeAll, or bisection if MagmaRangeV or MagmaRangeI.

For real matrices, only divide-and-conquer is available:
magma_[sd]syevd*

BTW, MAGMA is at version 2.5, not 2.0.

-mark

Post Reply