Thanks for your answer Richard,
I added abs_xsec_per_speciesAddLines into the cross-section agenda and now it
seems to work. But I do not know why lookupCalc method is working on a single
processor, even if I launched ARTS with the option -n 15.
Mattia
Da: Richard Larsson
Inviato: martedì 8 febbraio 2022 11:00
A: Mattia Sabatini
Cc: arts_users.mi@lists.uni-hamburg.de
Oggetto: Re: [arts-users] ARTS 2.5.0 - absorption
Dear Mattia,
It looks like you are removing line calculations from your lookup calculations.
Could you try adding abs_xsec_per_speciesAddLines to your lookup cross-section
agenda? I think this will solve the problem.
We are currently in a transition in the 2.5-branch of moving away from
abs_xsec_agenda entirely. There are some lingering problems when using lookup
table calculations at this time. Mainly, since the line calculations should
happen in propmat_clearsky_agenda now, they are not part of any of the default
cross-section agendas as in the past. The lookup generation interface is
currently under active development, so this interface might change in the near
future.
With hope,
//Richard
Den fre 21 jan. 2022 kl 14:53 skrev Mattia Sabatini
<mattia.sabat...@artov.ismar.cnr.it>:
Dear all,
I have a question concerning the absorption calculation with ARTS 2.5.0. Here
are some details about the controlfile I wrote:
1D atmosphere, clear sky, 4 IR channels
VectorLinSpace( f_grid, 2.02456915e13, 4.2827494e13, 5e8 ) – i.e. from 7um to
14.8 um
abs_species = "H2O, H2O-SelfContCKDMT100, H2O-ForeignContCKDMT100", "O3”, "CO2,
CO2-CKDMT100", "N2O", "CH4", "CFC11-HXSEC", "CFC12-HXSEC", "HCFC22-HXSEC",
"CFC113-HXSEC", "CFC114-HXSEC".
abs_linesReadSpeciesSplitCatalog(abs_lines, "cat/") - from
arts-xml-data-2.4/spectroscopy
ReadXsecData(basename="coefficients_arts/") – from arts-crossfit-main/
On-the-fly absorption set with
AgendaSet( propmat_clearsky_agenda ){
Ignore(rtp_mag)
Ignore(rtp_los)
Ignore(rtp_nlte)
propmat_clearskyInit
propmat_clearskyAddXsecAgenda
propmat_clearskyAddLines
propmat_clearskyAddHitranXsec
}
AgendaSet( abs_xsec_agenda ){
abs_xsec_per_speciesInit
abs_xsec_per_speciesAddConts
}
Batch_atm_fields_compact from Eresmaa 137L, gases VMR from FASCOD except for
CFC ones which are set as constant.
Surface set with
AgendaSet( surface_rtprop_agenda ){
specular_losCalc
Extract( surface_skin_t, t_surface_vector, ybatch_index )
surfaceFlatScalarReflectivity
}
(Surface_vector contains the surface skin temperatures from the Eresmaa surface
data. The correct skin temperature is extracted during ybatchCalc, accordingly
to ybatch_index)
This is my output for the first profile (channel BTs in Kelvin):
291.688825481684
288.903990987693
287.623301104291
284.402270055496
I repeated the calculation by using a lookup table, editing the controlfile:
AgendaSet( propmat_clearsky_agenda ){
Ignore(rtp_mag)
Ignore(rtp_los)
Ignore(rtp_nlte)
propmat_clearskyInit
propmat_clearskyAddFromLookup
propmat_clearskyAddHitranXsec
}
abs_lookupSetupBatch
abs_xsec_agenda_checkedCalc
lbl_checkedCalc
abs_lookupCalc
WriteXML( "binary", abs_lookup )
ReadXML( abs_lookup )
abs_lookupAdapt
The lookup calculation is done in few minutes, and the output I get this time
is:
293.350716564285
291.452571755867
296.771709918824
292.567643470471
Why are the two outputs for the same profile so different? Also, why is the
lookup table calculation so fast compared to version 2.4 (a couple of hours
with same f_grid and atmospheric profiles)? Where did I go wrong?
Mattia
_______________________________________________
arts_users.mi mailing list
arts_users.mi@lists.uni-hamburg.de
https://mailman.rrz.uni-hamburg.de/mailman/listinfo/arts_users.mi
_______________________________________________
arts_users.mi mailing list
arts_users.mi@lists.uni-hamburg.de
https://mailman.rrz.uni-hamburg.de/mailman/listinfo/arts_users.mi
