Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 07:12
    gagan-aryan commented #287
  • 07:06
    gagan-aryan commented #287
  • 07:05
    gagan-aryan commented #287
  • 06:47
    gagan-aryan synchronize #287
  • 06:43
    gagan-aryan synchronize #287
  • Jun 16 21:16
    anandxkumar commented #286
  • Jun 16 21:02
    anandxkumar synchronize #286
  • Jun 16 20:24
    erwanp commented #287
  • Jun 16 17:49
    gagan-aryan edited #287
  • Jun 16 17:26
    erwanp commented #284
  • Jun 16 17:17
    erwanp milestoned #287
  • Jun 16 16:17
    gagan-aryan commented #287
  • Jun 16 16:07
    gagan-aryan review_requested #287
  • Jun 16 16:07
    gagan-aryan review_requested #287
  • Jun 16 16:06
    gagan-aryan converted_to_draft #287
  • Jun 16 16:06
    gagan-aryan commented #287
  • Jun 16 16:02
    gagan-aryan opened #287
  • Jun 16 11:19
    erwanp edited #281
  • Jun 15 12:24
    gagan-aryan added as member
  • Jun 15 11:43
    anandxkumar added as member
Dirk van den Bekerom
@dcmvdbekerom
?
oops nevermind I'm having other problems
Dirk van den Bekerom
@dcmvdbekerom
ok so the problem is that the wavenumber axis does not have a constant spacing
this is technically true since it is off by a fraction of 1e-8 due to truncation errors
it seems for this reason it doesn't let me plot a spectrum with slit
by the way is this chat the right place to post this or is it better to start an issue on Github for such cases?
Erwan Pannier
@erwanp
If there's an error, you may want to start an Issue
You can resample the Spectrum, that may help fix the padding.
Erwan Pannier
@erwanp

or perhaps experimental_spectrum would not be needed at all (other than legacy perhaps)

That could be an Issue too, btw!
Did you also see this: https://radis.readthedocs.io/en/latest/spectrum/spectrum.html#initialize-from-text-file-or-arrays ?

Pankaj Mishra
@pkj-m
Hi all, I was working on implementing astropy units inplace of pint throughout RADIS and came across this particular test which calls the apply_slit(..norm_by="max") method. When it does so, the code tries to process a string "I/I0" as a unit. I have been looking all over the pint documentation but I still haven't been able to identify how this worked. Any leads on how this worked would be appreciated!
Oh wow, turns out the example Erwan shared just before my message is doing the very same thing!
Pankaj Mishra
@pkj-m
Just a guess but is that supposed to be a placeholder for representing a dimensionless quantity?
Erwan Pannier
@erwanp
yes! We could remove it. It's nice in the plots, but no need to make it appear as a unit now that Astropy is going to parse everything!
Pankaj Mishra
@pkj-m
So the BaseFactory class has a units' dictionary where this string appears as a value multiple times, for instance against keys : transmittance, transmittance_noslit and also as -ln(I/IO) for key absorbance. Can I simply replace all these by 1?
Dirk van den Bekerom
@dcmvdbekerom

or perhaps experimental_spectrum would not be needed at all (other than legacy perhaps)

That could be an Issue too, btw!
Did you also see this: https://radis.readthedocs.io/en/latest/spectrum/spectrum.html#initialize-from-text-file-or-arrays ?

Not yet, looks useful!

Erwan Pannier
@erwanp
@pkj-m yes. The only problem you'll get is when comparing against old spectra stored on disk as .spec with the unit as I/I0. There is a function in tools/database/load_spec() that can be used to check this deprecated syntax and update the unit to 1 as you load them.
Pankaj Mishra
@pkj-m
Hi @erwanp , do you think there might be any other ways of handling the issue of using strings like I/IO, -ln(I/IO) and eps in conjunction with Astropy? I have been at it for quite some time and it appears simply changing all occurrences of these strings to 1 won't resolve the issue. I have been forced to add a lot of if-else blocks in a lot of places to handle issues arising from the way astropy parses strings. For instance, while we could simply multiply two units represented as strings in pint by concatenating them with a * in between, in Astropy we need to first check whether both the strings are pure units or is one of them 1 (like in our case, following which we need to ensure that this string appears before the other string, nm in this case).
Inspite of making these changes, there are still other tests which break due to these changes, especially coming from matplotlib.
Erwan Pannier
@erwanp

simply changing all occurrences of these strings to 1 won't resolve the issue.

Can you post an example on the PR where it creates a problem? I can only think of the apply_slit() function.

Pankaj Mishra
@pkj-m
I have shared the console output on the PR ( #106 ) obtained from executing test_slit() locally on my laptop. The traceback seems to be hinting at something going wrong with matplotlib but I still not clear about it.
Erwan Pannier
@erwanp
@pkj-m I answered some of your problems: check the PR. the problem does not originate from fix_style but by how the LaTeX of the label is written.
Also, regarding the many print() comments in your code: I usually debug with the "%debug" feature of ipython, going through the different functions with "u" [up] and "d" [down], and evaluating the variables there. Of course it's up to you!
Pankaj Mishra
@pkj-m
Thanks for the help Erwan! I have implemented the changes you suggested. Right now I am trying to fix a few other errors that are coming up but hopefully this should be ready soon!
Pankaj Mishra
@pkj-m
Also, thanks for sharing the %debug feature. I was familiar with the gdb debugger for C++ but never bothered to find an equivalent for Python. I'll look into it!
Erwan Pannier
@erwanp
it's a very good one, and get all the power of having an interpreted language!
Erwan Pannier
@erwanp
Hey @pkj-m and @adityabharadwaj198 : I checked and merged your PRs, sorry for delay!!
aditya
@adityabharadwaj198
No problem! :)
Pankaj Mishra
@pkj-m
Looking good!
Dirk van den Bekerom
@dcmvdbekerom
Is there a diagram like this https://radis.readthedocs.io/en/latest/_images/RADIS_flow_chart.svg that includes the functions and modules that calculate each of the blocks
Pankaj Mishra
@pkj-m
Hey @erwanp @dcmvdbekerom and Minesi, thank you for the opportunity! I am really happy and extremely excited to be a part of RADIS! I really hope we can achieve a lot over the summer and make this project a big success!
Dirk van den Bekerom
@dcmvdbekerom
Congratulations!! I look forward to working with you on this project!
Erwan Pannier
@erwanp
Hey people! It seems that we only got one of the two project slots. It would have been even better if we got the two projects all together as they were quite synergistic. But let's see the good side of it : one is already good, and it's nice to have you onboard Pankaj!
Pankaj Mishra
@pkj-m
Thanks Erwan! Looking forward to a great summer with RADIS :)
Erwan Pannier
@erwanp
Hello @pkj-m @dcmvdbekerom : there are the Slack access : https://join.slack.com/t/radis-radiation/shared_invite/zt-ebxf872e-OQ5qGuf_HRMXTK9IXU6i3Q
Erwan Pannier
@erwanp
Permanent link to the same Slack, where most things happen now ! https://radis.github.io/slack-invite/
Stuart Mumford
@Cadair
Hello from Gitter
Cadair
@cadair:cadair.com
[m]
Hello from matrix
1 reply
erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]
works perfectly !
Cadair
@cadair:cadair.com
[m]
🥳
The users on slack don't render perfectly on gitter, but good enough.
Adarsh Desai
@slack_radis-radiation_U01T3FW977V:openastronomy.org
[m]

This message is replying to a Matrix event but we were unable to find associated bridged Gitter message to put it in the appropriate threaded conversation.

Oh! cool stuff

erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]

PR https://github.com/radis/radis/pull/271|#271 by Anand Kumar was just merged on the develop branch. It introduces an automatic scaling of the wavenumber grid, which adapts to the pressure, temperature and species considered. It is the best trade-off between accuracy and performance for a single spectrum, so I encourage you to try it. wstep='auto'

Will be available in the public 0.9.30 version incoming next week. And may become the default parameter in future versions if feedbacks are good.

4 replies
erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]
https://twitter.com/radis_radiation|@radis_radiation: Resolving 1.8M lines of the HITEMP CO2 database, live & without truncation, with GPU acceleration & the LDM algorithm ! @gully_ @pmxkcd @hitran https://t.co/nVQ5KeCad7 https://t.co/mYZsEnTH23
Dirk van den Bekerom
@slack_radis-radiation_U013V2RNQ10:openastronomy.org
[m]
...So if you will see a lot of new people here these coming days (hopefully!) that's the reason 🙂
erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]
Yes! !! 😛
gully
@slack_radis-radiation_U01G9J6A402:openastronomy.org
[m]
📺 Watching now 🍿 😄
gully
@slack_radis-radiation_U01G9J6A402:openastronomy.org
[m]
Nice talk! I understand the technique. It seems so clearly like the right approach. I am not sure how one can improve upon it? The main thing I can think of is the derivatives, either via autodiff or direct closed form evaluation.
Minou
@slack_radis-radiation_U018GDQHDK4:openastronomy.org
[m]
Hi guys,
How do you display the linesurvey module in a normal python code? As opposed to a jupyter notebook
from radis import calc_spectrum s = calc_spectrum( wavenum_min=2007, wavenum_max=2009, mole_fraction=0.1, path_length=10, # cm Tgas=700, molecule="CO", isotope=[1,2,3], databank="hitran", export_lines=True, ) s.line_survey(overlay="absorbance", barwidth=0.01) s.plot('absorbance')
1 reply
erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]
Finding of the day : package for post-processing of NMR/Infrared spectra : https://www.spectrochempy.fr/latest/index.html Not what RADIS does, but it's good to share all spectroscopy tools anyway !
erwan
@slack_radis-radiation_U9E3FCFDH:openastronomy.org
[m]
Quick update of the Website : https://radis.github.io/
• direct link to https://radis.app
• update links to https://radis.readthedocs.io/en/latest/auto_examples/index.html|Examples, and https://radis.github.io/radis-lab|Radis-Lab (online with pre-installed databases)
• mention the <https://www.sciencedirect.com/science/article/pii/S0022407320310049|LDM article>
• update features with link to https://user-images.githubusercontent.com/16088743/121789972-f34ab280-cbda-11eb-99b0-d7ba019d80d5.mp4|GPU-video ; mention of HITEMP auto download
Feedback welcome !
1 reply