Merge Parabands WFNs

[Austyn Masuno]

Hello BGW developers,

I was wondering if there is a way to merge WFNs that were created using Parabands? There is not much documentation online about this. I have run into an issue with needing to compute a lot of empty bands to converge GW calculations, hence why I used Parabands in the first place. However, I had split these up over kpts with the plan of eventually merging them back into one file, like how we usually can do with the output WFNs of pw2bgw. For the calculation I used the ELPA solver, block_size 64, and with 50 nodes, it takes about 1.5 hrs per kpt. The system has a Hamiltonian size 148702 and I had 72 kpts total. If merging the WFNs from Parabands is not possible, my next plan would be to just use less kpts so I don't have to split the calculation up, however this is not ideal. Please let me know if you have any advice. 

[Zhenglu Li]

Dear Austyn,

It should be the same procedure to merge WFNs.  Once in the BGW WFN format, the wfn_merge utility can process them.

Best,

Zhenglu

---

From: Austyn Masuno <austynma...@gmail.com>
Sent: Friday, October 3, 2025 6:46 AM
To: BerkeleyGW Help <he...@berkeleygw.org>
Subject: Merge Parabands WFNs

 

Hello BGW developers, I was wondering if there is a way to merge WFNs that were created using Parabands? There is not much documentation online about this. I have run into an issue with needing to compute a lot of empty bands to converge GW

Hello BGW developers,

I was wondering if there is a way to merge WFNs that were created using Parabands? There is not much documentation online about this. I have run into an issue with needing to compute a lot of empty bands to converge GW calculations, hence why I used Parabands in the first place. However, I had split these up over kpts with the plan of eventually merging them back into one file, like how we usually can do with the output WFNs of pw2bgw. For the calculation I used the ELPA solver, block_size 64, and with 50 nodes, it takes about 1.5 hrs per kpt. The system has a Hamiltonian size 148702 and I had 72 kpts total. If merging the WFNs from Parabands is not possible, my next plan would be to just use less kpts so I don't have to split the calculation up, however this is not ideal. Please let me know if you have any advice. 

--
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/46aae881-80c6-45b3-9819-a8ac77e790c6n%40berkeleygw.org.

[Austyn Masuno]

Hello Zhenglu,

Thank you for getting back to me. I have tried the same procedure, however, using wfn_merge utility on the output WFNs from Parabands gives this error:

ERROR: Failed read operation in read_binary_header on header record in mode GET.

Additionally, trying to inspect the output WFNs from Parabands using wfn_rho_vxc_info.x I get this error:

ERROR: Read unknown flavor '?<8f>¦ÂªN^E' in read_binary_header: must be 'Real'/'Complex'

Please let me know if you need any more information on my input/output files and what you think may be the issue.

Best,

Austyn

[Zhenglu Li]

Hi Austyn,

I assume you are dealing with WFN.h5, in HDF5 format?  If so, you can convert them into binary WFN files using hdf2wfn.x utility first, and then merge them.

Best,

Zhenglu

---

From: Austyn Masuno <austynma...@gmail.com>
Sent: Sunday, October 5, 2025 3:40 PM
To: BerkeleyGW Help <he...@berkeleygw.org>
Cc: Zhenglu Li <zhen...@usc.edu>; Austyn Masuno <austynma...@gmail.com>
Subject: Re: Merge Parabands WFNs

 

[Austyn Masuno]

Hi Zhenglu,

I see, that worked -- thank you!

Best,

Austyn