Comments (39)
@egawiser @janewman-pitt-edu @yymao @rmandelb Eric G./Jeff N. suggested doing a validation test based on equivalent-width (EW) distributions. However, the Galacticus output has line strengths rather than EWs and the SED supplied is probably too coarse grained to estimate EWs accurately (although it could be tried). Are there other tests based on line strengths that we could do? What applicable validation data is available?
from descqa.
It should be feasible to validate versus the H alpha, [O II], and [O III] luminosity functions. Adam Broussard has references assembled for H alpha, has been tasked with checking for [O III], and could add [O II] to that list. Note that for our specific LSS project, getting high enough line luminosities is actually what matters (not actually high EWs), so this is an entirely acceptable validation for that purpose.
from descqa.
Getting EWs right is a better check that you're giving the correct fluxes to the correct objects, though... coarse spectrum is probably fine for EW.
Any existing [OIII] or H alpha luminosity functions for z~1 will be for tiny areas with huge cosmic variance, so that doesn't strike me as a great option. (There is an [OII] luminosity function from DEEP2 -- IIRC Johan Comparat did that -- but the EWs are better measured in the data anyways, as they are not affected by flux calibrations).
from descqa.
Hi all, I have recently gotten a NERSC account so that I can access the protoDC2 catalog. I'm looking at the available quantities and it looks like I can access the band magnitudes, but not the emission line strengths. How can I get these added in?
from descqa.
These are native quantities, available in the catalog but not yet exposed in the reader. Please see https://github.com/LSSTDESC/gcr-catalogs for instructions on how to use the GCR (you need the add_quantities method) and https://portal.nersc.gov/project/lsst/descqa/v2/?run=2017-11-27_4&test=ListQuantities for a list of the native quantities (click on native_quantities.txt under proto-dc2_v2.0)
from descqa.
@adam-broussard get_quantities()
also accept native quantities.
@evevkovacs what's the native quantity names for the emission line strengths?
from descqa.
@adam-broussard Examples are: (see also my comment above for helpful links)
emissionLines/diskLineLuminosity:balmerAlpha6563:rest
emissionLines/diskLineLuminosity:balmerAlpha6563:rest:contam_nitrogenII6584
from descqa.
@adam-broussard @evevkovacs see usage on the updated demo notebook (search for "accessing native quantities" subsection)
from descqa.
Thanks Yao for the quick turn-around on this.
from descqa.
Thanks a bunch for the help so far - I have been able to access the values, but I can't figure out the units. It looks like the values for "emissionLines/totalLineLuminosity:balmerAlpha6563:rest" are around unity though. Can anyone help with what the native units are or where to track them down?
from descqa.
@adam-broussard Sorry for the delay in replying. We will have a units attribute soon, but in the mean time the units for luminosity (AB system) are 4.4659e13 W/Hz.
from descqa.
@evevkovacs I see that the units are L_nu - can you tell me how I could convert this to a true luminosity? I assume that the wavelength bin size on the SED should work as a differential nu to integrate it, and if so what is it? Thank you again!
from descqa.
Hi guys I hope everyone had a good winter break. I wanted to check in again now that we're on the other side of the new year. @evevkovacs @abensonca can you point me in the right direction for converting the emission lines from L_nu to L? Thanks in advance.
from descqa.
For the emission lines the quantities are already L, not L_nu. Eve's comment is for broad-band stellar continuum luminosities (which are L_nu in units of 4.4659e13 W/Hz). The emission lines are in units of Solar luminosities (3.839e26 W). So, to get equivalent widths you should just need to convert the continuum L_nu to L_lambda, and then take the ratio.
from descqa.
Great, thanks for the quick response!
from descqa.
@adam-broussard - have you been able to successfully access the emission line data? If you are actively working on this, perhaps you could summarize the status of what you are doing? In the issue page, I have not yet seen a concrete suggestion for how the validation test will be carried out, what validation dataset will be used, etc.
from descqa.
Hi @rmandelb ! I apologize for the late response - I was preoccupied with a proposal which was due at the end of last week and had to stop work on this project for a bit. We have some ideas of plotting the equivalent width distribution and comparing to observations, but nothing is solidly formed just yet. Give me a few days this week to get everything in a row and I'll report back in by Thursday with an update and a more concrete plan.
from descqa.
@evevkovacs Eric emailed me to say that I could pull the SED data from the GCR, and that they are named according to 'SED_l1_l2' where l1 and l2 are the starting and ending wavelengths respectively. I am seeing values such as 'sed_1000_246' - is it that this is instead the starting wavelength and the width (or something similar)? Also, can you tell me what the units are on these SED fluxes? Thanks!
from descqa.
@adam-broussard l2 is the width of the tophat filter, not the ending wavelength.
For units see Andrew Benson's comment above: "The emission lines are in units of Solar luminosities (3.839e26 W). So, to get equivalent widths you should just need to convert the continuum L_nu to L_lambda, and then take the ratio."
from descqa.
@evevkovacs great - does that make l1 the central wavelength of the tophat filter, or was I correct that it's the starting wavelength? And I think Adam was asking what kind of L_nu units the continuum values are in.
from descqa.
it's the start and width.
from descqa.
L_nu is in units of 4.4659e13 W/Hz
from descqa.
Awesome. Thank you all!
from descqa.
That is indeed what I was looking for - thank you for the clarificaiton, Eric! It is also useful to know that these are L_nu rather than F_nu as I probably would have assumed that they were fluxes naively. I will work on this and have an equivalent width distribution posted tonight or tomorrow morning.
from descqa.
Here is what I have found for the equivalent width distribution around a redshift of 1 (although I can do it for others). I am still searching out a paper which plots the equivalent width distribution for some redshift range. The closest I have been able to find is Marmo-Queralto et al. (2016) which deals with the average H-alpha equivalent width as a function of redshift.
from descqa.
This is a great start. Can you make the same plot for protoDC2 (eg pick z> .9) for comparison?
from descqa.
I did so, and it looks almost identical:
from descqa.
I think there's ~0% chance [OIII] should generally be as strong as H alpha. The plot looks suspicious to me... E.g. in Moustakas et al. 2005, H beta fluxes are typical ~1/3 as large as H alpha, and [OIII]/H beta ratios span ~1/3 to 3x .
from descqa.
@janewman-pitt-edu thanks for processing this through your well-trained neural network! I think the paper you mean is this one? http://adsabs.harvard.edu/abs/2006ApJ...642..775M That's low-redshift but does claim agreement with some z=1 observations. We know that [O III] gets much stronger at high redshift vs. [O II], and that should means it gets somewhat stronger vs. Ha, as reflected in this z=1 plot. In any case, hopefully @adam-broussard can generate the same plots for z=0 (say z<0.2) and z=0.5, where there's more literature to compare with.
from descqa.
There's not much evolution in BPT diagrams by z~1 (especially for L* galaxies) -- things shift to lower metallicity at higher redshifts than that (the [OII] EW distribution is pretty invariant from redshift at z=0.8-1.4 as I recall, by the way).
I'd suggest looking at histograms of [OIII]/H alpha, [OIII]/H beta, and [OIII]/[OII]; that should show immediately if [OIII] is too tied to H alpha, for instance.
The simulations are somewhat incomplete at z<0.2, so z~0.5 would be probably a better place to do the comparison.
from descqa.
PS Yes, that's the right paper. I found it via arxiv where the date was 2005.
from descqa.
In case it's helpful, I've compiled plots for the [OII]/[OIII]/H beta flux ratios for DEEP2 galaxies (there's a limited redshift range where we can get all of these lines), as well as the [OII] EW distribution. I've produced plots of the ratios, etc. as a function of z (where dependence is weak and may just be due to the R=24.1 limit corresponding to more luminous objects at higher redshifts), as well as histograms. All of these are quantities you should be able to compare against (with the same R=24.1 cut applied).
(PS I'm applying a 3 sigma detection cut in [OII] EW for the plots with [OII], or in H beta EW for the [OIII]/Hb ratio).
from descqa.
I have generated the plot for z = 0.5 but unfortunately I am getting the same results. I have also included a histogram showing the distribution of different equivalent width ratios, which seems to show that [OIII]/[OII] and H-alpha/[OII] EW is approximately 10, but H-alpha/[OIII] EW is approximately 1.
from descqa.
Yeah, that's definitely not right. [OIII] should generally be less than [OII], not 10x higher, based on the real data.
from descqa.
@adam-broussard I see that you have made much progress on this. Do you think it is useful to create a test in DESCQA, basically porting what you have done into the framework? I can offer help, of course.
Also, how should we validate the results? @janewman-pitt-edu mentioned some potential validation data from DEEP2. Should we use them to compare with the catalog result when we implement this test in DESCQA? @egawiser @adam-broussard.
from descqa.
@yymao Eric and I talked today and we think it would be a good idea to put this in DESCQA. We discussed some possible tests and came up with:
- Fitting the exponential decline on the right side of the equivalent width distribution
- The ratio of primary and secondary modes (essentially the modes of equivalent widths of the star forming vs quiescent population)
- The equivalent width of the mode
from descqa.
@janewman-pitt-edu could you send me the data you used to generate those example plots? That way I can easily compare between the data and protoDC2 as well as potentially use them for the tests I outlined above.
from descqa.
@adam-broussard sounds like a good plan. Let me know how I can help you putting this test into the framework!
from descqa.
Sure, send me a direct email to [email protected] and I can reply to it with files.
Best,
Jeff
from descqa.
Related Issues (20)
- Tree ring test
- validate instance catalogs to filter out offending AGN HOT 3
- number counts test updates HOT 6
- Update DESCQA web app's landing page HOT 4
- Ability to "tag" a run after the run is complete
- `sklearn.cluster.k_means` not working when `n_jobs=-1` is set HOT 1
- README is pointing to jupyter-dev
- Segmentation fault running some correlation function tests HOT 2
- Shear Test fails due to OSError: libgfortran.so.3: cannot open shared object file: No such file or directory HOT 7
- shear_test fails due to camb attribute error
- Move from project/projecta to cfs
- Python Environment Name Change stack => desc HOT 1
- Need to update versions of gsl and cray-fftw in run_master.sh HOT 2
- Set HDF5_USE_FILE_LOCKING=FALSE in run_master.sh
- Make DESCQA compatible with new desc-python environment HOT 6
- Is there a way to pass `external_data_dir` as an argument? HOT 5
- Revert to using the desc-python env at NERSC
- Experiment with generalizing tests beyond GCRCatalogs HOT 3
- Versioning of releases and plans any plans for an updated release? HOT 1
- DeltaSigma test failures HOT 1
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from descqa.