Dear ALPS Users and Developers,
I have been observing some unexpected behavior when using translation symmetry with the ALPS fulldiag application. The detailed description (PDF), as well as a simple example and the output in question (.zip), are attached.
More recently, I have discovered another issue: If I consider a single spin-1/2 in the Sz basis subjected to magnetic field pointing along the y-direction, the eigenvectors and energies are calculated correctly but the expectation value of e.g. the identity-operator is zero. I have tracked down the problem to a surplus complex conjugation in the calculation of the scalar product of complex vectors and the attached patch seems to fix it. Moreover, the originally encountered problem (cf. first paragraph) seems to be no more than a different manifestation of the same bug.
I hope this information will be useful.
Best regards,
Marek Pikulski Laboratory for Solid State Physics ETH Zurich
Dear Marek,
Thank you very much for finding this bug. I have applied the patch and hope that it will soon propagate to the nightly snapshot. Does this also fix all of your original problem?
Matthias
On 13 Apr 2015, at 16:01, Pikulski Marek marekp@student.ethz.ch wrote:
Dear ALPS Users and Developers,
I have been observing some unexpected behavior when using translation symmetry with the ALPS fulldiag application. The detailed description (PDF), as well as a simple example and the output in question (.zip), are attached.
More recently, I have discovered another issue: If I consider a single spin-1/2 in the Sz basis subjected to magnetic field pointing along the y-direction, the eigenvectors and energies are calculated correctly but the expectation value of e.g. the identity-operator is zero. I have tracked down the problem to a surplus complex conjugation in the calculation of the scalar product of complex vectors and the attached patch seems to fix it. Moreover, the originally encountered problem (cf. first paragraph) seems to be no more than a different manifestation of the same bug.
I hope this information will be useful.
Best regards,
Marek Pikulski Laboratory for Solid State Physics ETH Zurich
<Question_files.zip><alps_r7462_fulldiag_conj_patch.diff><Question.pdf>
Dear Matthias,
As far as I can tell at the moment, yes.
Best regards,
Marek
On 13/04/15 21:29, "Matthias Troyer" troyer@phys.ethz.ch wrote:
Dear Marek,
Thank you very much for finding this bug. I have applied the patch and hope that it will soon propagate to the nightly snapshot. Does this also fix all of your original problem?
Matthias
On 13 Apr 2015, at 16:01, Pikulski Marek marekp@student.ethz.ch wrote:
Dear ALPS Users and Developers,
I have been observing some unexpected behavior when using translation symmetry with the ALPS fulldiag application. The detailed description (PDF), as well as a simple example and the output in question (.zip), are attached.
More recently, I have discovered another issue: If I consider a single spin-1/2 in the Sz basis subjected to magnetic field pointing along the y-direction, the eigenvectors and energies are calculated correctly but the expectation value of e.g. the identity-operator is zero. I have tracked down the problem to a surplus complex conjugation in the calculation of the scalar product of complex vectors and the attached patch seems to fix it. Moreover, the originally encountered problem (cf. first paragraph) seems to be no more than a different manifestation of the same bug.
I hope this information will be useful.
Best regards,
Marek Pikulski Laboratory for Solid State Physics ETH Zurich
<Question_files.zip><alps_r7462_fulldiag_conj_patch.diff><Question.pdf>
comp-phys-alps-users@lists.phys.ethz.ch