Further, I just literally noticed yesterday that in case siesta would pass the stress tensor to do a NPT simulation, this also has to be fixed.
I did not yet have time to test is, but essentially the line ~275-276 in iosockets.F90:
! Find virial tensor for i-pi, in master's units
vir(:) = -s * cellv * fdf_convfac( siesta_xunit, master_xunit )**3
should also be exchanged by
! Find virial tensor for i-pi, in master's units
vir(:) = -TRANSPOSE(s) * cellv * fdf_convfac( siesta_xunit, master_xunit )**3
I think it is worth implementing this fix even without further testing as it is the only way everything is consistent.
Cheers and many thanks,
Mariana
> On 29. Jul 2019, at 11:16, Nick Papior <email address hidden> wrote:
>
> ** Changed in: siesta/4.0
> Status: In Progress => Fix Committed
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1835196
>
> Title:
> Cell should be the transpose when communicating between siesta and
> i-PI
>
> Status in Siesta:
> In Progress
> Status in Siesta 4.0 series:
> Fix Committed
> Status in Siesta 4.1 series:
> In Progress
>
> Bug description:
> Because of the matrix definitions in i-PI and SIESTA, currently the
> cell that is being passed to SIESTA is the transpose of what it should
> be. This only affects non-orthogonal cells. The following fix in the
> routine iosockets.F90 seems to fix the problem in my tests:
>
> From line 197:
>
> cell = RESHAPE( c, (/3,3/) )
> + cell = TRANSPOSE(cell)
>
> That's all.
>
> Cheers,
>
> Mariana
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/siesta/+bug/1835196/+subscriptions
---------------------------------------------------------------
Dr. Mariana Rossi e-mail: <email address hidden> <mailto:<email address hidden>>.de
Faradayweg 4-6 phone: +49 30 84134839
Theory Department
14195 Berlin
Germany
Hi Nick,
Further, I just literally noticed yesterday that in case siesta would pass the stress tensor to do a NPT simulation, this also has to be fixed.
I did not yet have time to test is, but essentially the line ~275-276 in iosockets.F90:
! Find virial tensor for i-pi, in master's units
vir(:) = -s * cellv * fdf_convfac( siesta_xunit, master_xunit )**3
should also be exchanged by
! Find virial tensor for i-pi, in master's units
vir(:) = -TRANSPOSE(s) * cellv * fdf_convfac( siesta_xunit, master_xunit )**3
I think it is worth implementing this fix even without further testing as it is the only way everything is consistent.
Cheers and many thanks,
Mariana
> On 29. Jul 2019, at 11:16, Nick Papior <email address hidden> wrote: /bugs.launchpad .net/bugs/ 1835196 /bugs.launchpad .net/siesta/ +bug/1835196/ +subscriptions
>
> ** Changed in: siesta/4.0
> Status: In Progress => Fix Committed
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https:/
>
> Title:
> Cell should be the transpose when communicating between siesta and
> i-PI
>
> Status in Siesta:
> In Progress
> Status in Siesta 4.0 series:
> Fix Committed
> Status in Siesta 4.1 series:
> In Progress
>
> Bug description:
> Because of the matrix definitions in i-PI and SIESTA, currently the
> cell that is being passed to SIESTA is the transpose of what it should
> be. This only affects non-orthogonal cells. The following fix in the
> routine iosockets.F90 seems to fix the problem in my tests:
>
> From line 197:
>
> cell = RESHAPE( c, (/3,3/) )
> + cell = TRANSPOSE(cell)
>
> That's all.
>
> Cheers,
>
> Mariana
>
> To manage notifications about this bug go to:
> https:/
------- ------- ------- ------- ------- ------- ------- ------- -------
Dr. Mariana Rossi e-mail: <email address hidden> <mailto:<email address hidden>>.de
Faradayweg 4-6 phone: +49 30 84134839
Theory Department
14195 Berlin
Germany
http:// th.fhi- berlin. mpg.de/ groups/ sabia/ <http:// th.fhi- berlin. mpg.de/ groups/ sabia/> www.fhi- berlin. mpg.de/ ~rossi/> 0735-11600- 23.05:fhi --------------
<http://
-------- key:1-0.