Eigenvalue problem

[z_wang]

Dear Steve,

I have a program to run eigenvalue calculation subrountine EVCRG hundreds of times.

But at some run (after 200 times) it reports 'unsolved exception'. Though I have used the RESET(0, 1, 0) to prevent this happening, but it failed. Curious enough, the code on calculating the right eigenvector succeeds, but the left eigenvector calculating fails. Do you have any idea on that? If possible, I'd like to email my codes to you, but I don't know your email address. If you like, youcould contact me via z.wang@polyu.edu.hk. Thank you very much.

-----------------------------------------------------------------------------

! Turn on printing and turn off stopping for all error types.
CALL ERSET (0, 1, 0)
CALL EVCRG (FULL_A, EVAL, R_EVEC)! right eigenvector, succeed

! ------error handling-------
ICODE = IERCD()
ITYPE = N1RTY(1)
IF (ICODE == 1 .OR. ITYPE==4 .OR. ITYPE==5) THEN
HAS_ERROR = 1
RETURN
END IF
! -----------------
FULL_A=TRANSPOSE(FULL_A)
CALL ERSET (0, 1, 0)
CALL EVCRG (FULL_A, EVAL, L_EVEC)! left eigenvector, failed

! ------error handling-------
ICODE = IERCD()
ITYPE = N1RTY(1)
IF (ICODE == 1 .OR. ITYPE==4 .OR. ITYPE==5) THEN
HAS_ERROR = 1
RETURN
END IF

---------------------------------------------------------------------------------

Best regards,

Wang Zhen, James

[Steven_L_Intel1]

I assume this is an IMSL routine? You can submit the problem to us through Intel Premier Support.

[John4]

I don't know if this can be of any help, but I've noticed that the IMSL routines tend to have an odd behavior when called many times, unless you provide an explicit workspace for them. In your case, the explicit interface is provided by E8CRG.

John.