A bug in iMPSOps_include.f90 makes the Fortran function ZNAUPD return -7, which crashes the simulation when computing observables.
The bug comes from the fact that lines 440-442 should be placed before line 436-438, otherwise the variable ncv on line 437 is not computed correctly resulting in workl not correctly allocated. The return value -7 means "insufficient storage allocated for private working arrays" (i.e. workl is too small).
Also, "dnaupd" and "dneupd" on lines 474 and 487 should be changed to "znaupd" and "zneupd", to make debugging easier.
Hello,
thank you for the detailed report. The corrections are necessary but are
not sufficient for the example with complex operators I have right now due
to a second error which follows. I have a branch and will update the ticket
in case I find something.
Sincerely,
Daniel
On Fri, Mar 6, 2020 at 9:37 PM Princehahaha princehahaha@users.sourceforge.net wrote:
Related
Tickets: #43