possible bug in plotxct
13 views
Skip to first unread message
Lai MingRui
Jun 6, 2025, 2:37:46 AMJun 6
to BerkeleyGW Help
Dear developers,
I've been trying to use the plotxct module in BerkeleyGW/3.0.1 and 4.0 and I encountered a recurring error. The error message is from line 170 in the PlotXct/input.f90 file when the program cannot find the corresponding kpoint in the WFN_fi file.
I noticed that this occurs when the program is trying to determine if two kpoints are the same, in line 159 of input.f90 where the variable kt is calculated. I encountered a case where although the two kpoints being checked are suppose to be the same, they differ by a small amount due to numerical errors, which is alright since there is a tolerance of 1e-4. The problem here is that for the kpoints in my WFN_fi and eigenvectors files the error is negative and it causes kt to be 0.9999... due to the use of the MOD function. The next line will then deem this two kpoints as different even though they are the same.
I'm wondering if this error has been encountered before. It seems like the checkbz routine shifts some kpoints by a G vector, because the kpoints in the WFN and eigenvectors file all have values in the range from 0 to 1.
Best Regards,
Lai MingRui
I've been trying to use the plotxct module in BerkeleyGW/3.0.1 and 4.0 and I encountered a recurring error. The error message is from line 170 in the PlotXct/input.f90 file when the program cannot find the corresponding kpoint in the WFN_fi file.
I noticed that this occurs when the program is trying to determine if two kpoints are the same, in line 159 of input.f90 where the variable kt is calculated. I encountered a case where although the two kpoints being checked are suppose to be the same, they differ by a small amount due to numerical errors, which is alright since there is a tolerance of 1e-4. The problem here is that for the kpoints in my WFN_fi and eigenvectors files the error is negative and it causes kt to be 0.9999... due to the use of the MOD function. The next line will then deem this two kpoints as different even though they are the same.
I'm wondering if this error has been encountered before. It seems like the checkbz routine shifts some kpoints by a G vector, because the kpoints in the WFN and eigenvectors file all have values in the range from 0 to 1.
Best Regards,
Lai MingRui
Zhenglu Li
Jun 6, 2025, 3:09:33 AMJun 6
to Lai MingRui, BerkeleyGW Help
Dear MingRui,
Looks like this comes from the accuracy in the kpoints, based on your description. 1e-4 seems too big, and different parts of the code may assume different thresholds.
Best,
Zhenglu
From: Lai MingRui <laiming...@gmail.com>
Sent: Thursday, June 5, 2025 11:37 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Subject: possible bug in plotxct
Sent: Thursday, June 5, 2025 11:37 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Subject: possible bug in plotxct
--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/e02089c9-d33f-4f4e-8ca3-28f0cc665a34n%40berkeleygw.org.
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/e02089c9-d33f-4f4e-8ca3-28f0cc665a34n%40berkeleygw.org.
Reply all
Reply to author
Forward
0 new messages