Faulty Output of "spin components" for deprecated SpinOrbit switch

[Peter Christian Schmitz]

VERSION: siesta-psml-R1--709-594

When the switches

SpinOrbit true
Spin non-polarized

are set together, the code prints

redata: Spin configuration = spin-orbit+onsite
redata: Number of spin components = 8

, suggesting a non-collinear calculation with onsite SOC. However, the real nonpolarized calculation with

# SpinOrbit false (not set)
Spin non-polarized

yields

redata: Spin configuration = none
redata: Number of spin components = 1

and the SAME energy, which is not possible if the first was true. Additionally, a calculation with off-site SOC

# SpinOrbit false (not set)
Spin spin-orbit

yields

redata: Spin configuration = spin-orbit+offsite
redata: Number of spin components = 8

So the presence of the supposed-to-be deprecated SpinOrbit causes an output which does not match the actually performed calculation.
Likely, the output routine still depends on SpinOrbit and needs to be updated.

[Nick Papior (nickpapior)]

Thanks for the bug-report!

This should already be fixed in the latest version of the PSML branch.

You could you try with:
https://bazaar.launchpad.net/~siesta-pseudos-bases/siesta/trunk-psml/tarball/598

I'll consider this resolved. :)

[Alberto Garcia (albertog)]

It is indeed a bug and it affects the trunk version also.
The fix is coming soon.

[Nick Papior (nickpapior)]

Ah, so something changed between 4.1 and trunk... I see...

For the record, 4.1 implicitly seems to have the fix, without warning messages.

[Alberto Garcia (albertog)]

Actually, 4.1 also exhibits the bug. I had assumed it was a 'trunk' issue and
started with that branch, instead of checking.
The problem is that the chain of if_then_else's acts on several logical variables that can
be true at the same time (e.g., spin%SO and spin%none)

[Alberto Garcia (albertog)]

I have now backported the fix to 4.1, and propagated it to trunk-psml.

In trunk and trunk-psml, the net effect of the bug was that "spin%SO" remained true, while
neither spin%SO_offsite nor spin%SO_onsite were set. Hence no SO contributions were really computed, and the energy was the same as the "spin%none" case.

[Zeila Zanolli (zeila)]

great! thanks!

 Z
---------------------------------------------------------------
Dr. Zeila Zanolli

Ramon y Cajal fellow at the
Institut Català de Nanociència i Nanotecnologia (ICN2), Barcelona, Spain
Steering Committee, European Theoretical Spectroscopy Facility
http://www.etsf.eu/
Board Member, Young Academy of Europe http://yacadeuro.org/

Phone: [phone]
Email: <email address hidden>
Web: http://zeilazanolli.wordpress.com/home
Twitter: @ZeilaZanolli
<https://vpn.icn2.cat/proxy/32d9bc05/http/intranet/Lists/Institutional%20signature/@ZeilaZanolli>
ResearcherID: http://www.researcherid.com/rid/F-9568-2010
----------------------------------------------------------------

On Sun, Jan 13, 2019 at 5:20 PM Alberto Garcia <email address hidden> wrote:

> I have now backported the fix to 4.1, and propagated it to trunk-psml.
>
> In trunk and trunk-psml, the net effect of the bug was that "spin%SO"
> remained true, while
> neither spin%SO_offsite nor spin%SO_onsite were set. Hence no SO
> contributions were really computed, and the energy was the same as the
> "spin%none" case.
>
>
> ** Changed in: siesta/4.1
> Status: New => Fix Committed
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1811464
>
> Title:
> Faulty Output of "spin components" for deprecated SpinOrbit switch
>
> Status in Siesta:
> Fix Committed
> Status in Siesta 4.1 series:
> Fix Committed
>
> Bug description:
> VERSION: siesta-psml-R1--709-594
>
> When the switches
>
> SpinOrbit true
> Spin non-polarized
>
> are set together, the code prints
>
> redata: Spin configuration = spin-orbit+onsite
> redata: Number of spin components = 8
>
> , suggesting a non-collinear calculation with onsite SOC. However, the
> real nonpolarized calculation with
>
> # SpinOrbit false (not set)
> Spin non-polarized
>
> yields
>
> redata: Spin configuration = none
> redata: Number of spin components = 1
>
> and the SAME energy, which is not possible if the first was true.
> Additionally, a calculation with off-site SOC
>
> # SpinOrbit false (not set)
> Spin spin-orbit
>
> yields
>
> redata: Spin configuration = spin-orbit+offsite
> redata: Number of spin components = 8
>
>
> So the presence of the supposed-to-be deprecated SpinOrbit causes an
> output which does not match the actually performed calculation.
> Likely, the output routine still depends on SpinOrbit and needs to be
> updated.
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/siesta/+bug/1811464/+subscriptions
>