Re: [arts-users] ARTS-3

2019-12-28 Thread Patrick Eriksson

Dear Eric,

Nice to hear that you have long term plans around your retrievals and 
that they involve ARTS.


Very good questions, for sure important information for ARTS users.


1.Will it be possible to continue using Qpack with the new release of 
ARTS and if yes, will it involve some major changes in the way of 
programming a new retrieval ?


Presently, Qpack is not working with the development version of ARTS. 
When ARTS gets a bit more stable, I will fix this. So yes, it will be 
possible to use Qpack with ARTS-3. (But this could be the last major 
upgrade of Qpack, as discussed below.)



2. What are (if existing) the long-term view of the ARTS developers 
regarding the use of Qpack with ARTS-3 ? Will it stay supported as the 
main way of interacting with ARTS or are there some inclinations towards 
some other ways (for instance using Typhon) ?


To avoid a possible confusion. Qpack is of restricted scope. In Matlab 
the interface to ARTS is called qarts.


When it comes to qarts vs. Typhon, that is  "the main way of interacting 
with ARTS", the answer is clearly Typhon.


If your question only refers to how to handle your retrievals, it is 
harder to give a firm answer. But in a longer perspective Typhon is 
still probably the safest choice.


The core developers have mainly moved to Python. I am probably the only 
one that still use Matlab. However, also I would prefer to use Python, 
as it's free software. I have just not found the time to make the switch 
(but now it seems that I must learn Python, to handle an upcoming course).


As OEM now is integrated into ARTS and the Python interface to ARTS is 
totally general, there are already less restrictions when using Python. 
(This has the consequence that you need to know ARTS quite well, for 
good and bad.)


On the other hand, Qpack is a dedicated retrieval tool and offers 
features that not yet are at hand in Typhon. For example, in Typhon you 
need to handle the data largely yourself, while in Qpack there is a 
standardized way to define a priori data, L1 and L2.


At least some of these features should be added to Typhon, but exactly 
how and when I can not say. This is mainly a matter of manpower. Both 
Simon and I are very busy. In addition, for some time we must focus on 
ARTS itself.


Sorry, if the answer is a bit vague, but it is hard to commit to things 
without dedicated funding. I hope you understand. Don't hesitate to come 
back if you have any specific question.


Bye,

Patrick







___
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] ARTS-3

2019-12-12 Thread eric.sauvageat
Hello,

My name is Eric Sauvageat and I am a new PhD student in the MW Physics group at 
the Institute of Applied Physics of University Bern.

My main task for the coming year it to set up a new harmonized retrieval for 
the two ground-based MW radiometers measuring Ozone profile in Switzerland.
In this regards, the goal is to continue using ARTS as radiative transfer model 
and switch to ARTS-3 when it will be released.

The current retrievals are using Matlab and Qpack for the interaction with ARTS 
and there is now the question arising if the new retrievals should be written 
on Matlab or switched to Python.
In order to take a decision within the group, I would like to ask you the 
following questions regarding the future views for ARTS-3:

1. Will it be possible to continue using Qpack with the new release of ARTS and 
if yes, will it involve some major changes in the way of programming a new 
retrieval ?

2. What are (if existing) the long-term view of the ARTS developers regarding 
the use of Qpack with ARTS-3 ? Will it stay supported as the main way of 
interacting with ARTS or are there some inclinations towards some other ways 
(for instance using Typhon) ?

I am looking forward to hearing from you and I thank you in advance for your 
answers.

Best regards,
Eric
___
arts_users.mi mailing list
arts_users.mi@lists.uni-hamburg.de
https://mailman.rrz.uni-hamburg.de/mailman/listinfo/arts_users.mi