topic Hi, in Intel® oneAPI Math Kernel Library

dfeast_sygv -4 error, BUT B IS POSITIVE DEFINITE!!!

yera__rolando — Wed, 13 Feb 2019 15:18:54 GMT

Hello all,

I am using the extended eigensolver routines specifically dfeast_sygv function, and I get the following error

==>INFO code =: -4
Intel MKL Extended Eigensolvers Error: Matrix B is not positive definite.

But the matrix B is positive definite!!!

B =

+2.222e-01      +0.000e+00      +1.111e-01      +0.000e+00      +1.111e-01      +0.000e+00      +5.556e-02      +0.000e+00
+0.000e+00      +2.222e-01      +0.000e+00      +1.111e-01      +0.000e+00      +1.111e-01      +0.000e+00      +5.556e-02
+1.111e-01      +0.000e+00      +2.222e-01      +0.000e+00      +5.556e-02      +0.000e+00      +1.111e-01      +0.000e+00
+0.000e+00      +1.111e-01      +0.000e+00      +2.222e-01      +0.000e+00      +5.556e-02      +0.000e+00      +1.111e-01
+1.111e-01      +0.000e+00      +5.556e-02      +0.000e+00      +2.222e-01      +0.000e+00      +1.111e-01      +0.000e+00
+0.000e+00      +1.111e-01      +0.000e+00      +5.556e-02      +0.000e+00      +2.222e-01      +0.000e+00      +1.111e-01
+5.556e-02      +0.000e+00      +1.111e-01      +0.000e+00      +1.111e-01      +0.000e+00      +2.222e-01      +0.000e+00
+0.000e+00      +5.556e-02      +0.000e+00      +1.111e-01      +0.000e+00      +1.111e-01      +0.000e+00      +2.222e-01

using dfeast_syev function give the following eigenvalues.

+5.556e-02
+5.556e-02
+1.667e-01
+1.667e-01
+1.667e-01
+1.667e-01
+5.000e-01
+5.000e-01

all positive.

Here go the matrix A for the general eigenvalues problem

A =

+4.569e+02      +0.000e+00      -3.046e+02      +0.000e+00      +7.616e+01      +0.000e+00      -2.285e+02      +0.000e+00
+0.000e+00      +4.569e+02      +0.000e+00      +7.616e+01      +0.000e+00      -3.046e+02      +0.000e+00      -2.285e+02
-3.046e+02      +0.000e+00      +4.569e+02      +0.000e+00      -2.285e+02      +0.000e+00      +7.616e+01      +0.000e+00
+0.000e+00      +7.616e+01      +0.000e+00      +4.569e+02      +0.000e+00      -2.285e+02      +0.000e+00      -3.046e+02
+7.616e+01      +0.000e+00      -2.285e+02      +0.000e+00      +4.569e+02      +0.000e+00      -3.046e+02      +0.000e+00
+0.000e+00      -3.046e+02      +0.000e+00      -2.285e+02      +0.000e+00      +4.569e+02      +0.000e+00      +7.616e+01
-2.285e+02      +0.000e+00      +7.616e+01      +0.000e+00      -3.046e+02      +0.000e+00      +4.569e+02      +0.000e+00
+0.000e+00      -2.285e+02      +0.000e+00      -3.046e+02      +0.000e+00      +7.616e+01      +0.000e+00      +4.569e+02

can anybody help me here???

I'm getting something wrong???

Im using linux (debian testing).

thanks in advance.

ps. Sorry about the english.

Which version of mkl do you

Gennady_F_Intel — Fri, 15 Feb 2019 07:19:00 GMT

Which version of mkl do you try?

Hi,

yera__rolando — Fri, 15 Feb 2019 14:06:18 GMT

Hi,

Im using MKL 2019 update 1 version, installed with parallel_studio_xe_2019_update1_cluster_edition package under student licence. Im also using the ICC version that comes in the package.

Ok, could you get us the C or

Gennady_F_Intel — Fri, 15 Feb 2019 14:09:52 GMT

Ok, could you get us the C or F reproducer to check this behavior on our side with the newest versions 2019.3?

Hi, Here go the src file.

yera__rolando — Fri, 15 Feb 2019 17:36:01 GMT

Hi,

Here go the src file.

Thanks for noticing this case

Gennady_F_Intel — Sat, 16 Feb 2019 03:40:23 GMT

Thanks for noticing this case. This looks like the issue which will be escalated.

Hi,

yera__rolando — Sat, 16 Feb 2019 14:10:59 GMT

Hi,

I switch to the csr version "dfeast_scsrgv" for the general eigenvalues problem and now I getting SIGSEGV error, the "dfeast_scsrev" works fine with B for the standard eigenvalues problem.

Here go the src file.

There was a bug in the code

yera__rolando — Sat, 16 Feb 2019 14:35:52 GMT

There was a bug in the code of the previous src file (conformation of the rowIndex data), here go the fixed version, still the behavior of the "dfeast_scsrgv" remain the same.

thanks for the update.

Gennady_F_Intel — Mon, 18 Feb 2019 03:48:37 GMT

thanks for the update.

the matrix checker has been

Gennady_F_Intel — Mon, 27 May 2019 09:51:27 GMT

the matrix checker has been fixed in the latest 2019 u4 version of mkl.

Thanks you very much!!!!

yera__rolando — Mon, 27 May 2019 14:42:36 GMT

Thanks you very much!!!!