-J option

[xjxiao]

Dear all,
I have two questions and would greatly appreciate your answers.

1. I am confused about the purpose of the `-J` option in Yambo. If I run `yambo -F hf_10Ry.in -J 3D` and generate the `3D` folder, then perform the following steps:
① Enter `$ cp hf_10Ry.in hf_20Ry.in`,
② Edit the file, changing `EXXRLvcs` from 10 Ry to 20 Ry,
③ Run `$ yambo -F hf_20Ry.in -J 3D`,
will Yambo read the data previously calculated with `EXXRLvcs=10 Ry` from the `3D` folder during execution? Could this lead to incorrect output results for the calculation with `EXXRLvcs=20 Ry`?

2. If the quasiparticle energies calculated using the same input file on two different servers differ by approximately 0.02 eV, is this error acceptable? Could this be caused by differences in my input files, or by different versions of Yambo (although both seem to be Yambo 5.1.0), or by other random factors?

Thanks!

Yours,
Xiao

[Daniele Varsano]

Dear Xiao,

1) when you add a -J dirname option, the logic is that yambo will read all the database present in the SAVE and in dirname. New databases will be stored in dirname.
So, the answer is yes, yambo will read the previously stored database. Anyway, it will also recognise that the database is calculated with a parameter different from the one indicated in the input file. In this case, depending on the entity, in the report it will either emit a WARNING, or an ERROR. In the case the error is emitted, a new database is recalculated with the parameters indicated in the input. Inspecting the report file, you can recognize what happened. Search for ndb. in the report and you will see which database has been read (RD) and written (WR).

In any case you can avoid to read the previous database by doing for example:
First run -J 3D_10Ry
Second run -J 3D_20 Ry

2) First of all, 5.1.0 is a quite old release, I strongly suggest you to update to a newer version. The energy different is quite small, anyway you can do a diff of the two output and see if there are differences in the parameters reported in the header. Next, if you are using random integration technique for the divergence of the coulomb potential, small differences can appear due to the stochastic integration.

Best,

Daniele