Problem reading star truth light curves in "stars/truth_star_variability_v1-0-0.parquet"


I am trying to read the truth values of lightcurves of objects with “MLT” model that include the microlensing events. I am running the " Read the DESC truth tables in parquet format" Jupyter Notebook at this link:

First, I read the “stars/truth_star_summary_v1-0-0.parquet” file and choose an object with “MLT” model and find its ID. (image1)

Then I use the ID to get the true lightcurve from the “stars/truth_star_variability_v1-0-0.parquet” file. It returns an empty dataframe. I have tried many MLT ids and all of them return empty dataframes. I also tried all of the MLT model IDs in the summary file and tried to read them from the variability file but it returned empty dataframes.

Hi Somayeh – This is interesting indeed. I see in the DC2 Data Release Note (Sec. 3.3) that

Variability Tables, one each for stars and SNe. These tables have a row for each observation of a variable source, excluding stars whose variability would not be detectable by LSST. Note that most stars are classed as “variable” but only about 10% vary enough to be included in the variability table; for such stars, the flag field above_threshold in the star summary table is set.

I thus tried modifying your query to tx = np.where((result_star_sum['model'][:] == 'MLT') & (result_star_sum['above_threshold'][:] == 1)), and indeed this returned no results (it does return results for “applyRRly” and “kplr” types, though. This must mean that the “MLT” objects were not included in the table. :frowning:

I will ask around a bit to see if I can find out whether there is a way to access those lightcurves.

Here is the response that I got from Jim Chiang re: accessing lightcurves for MLT objects:

The only way to do it that I know of would be to run the code that generated those light curves for the DC2 instance catalogs. I commented on that here – at the bottom of that comment is a link to the code in LSSTDESC. That code in turn depends on some lsst_sims code.

I’m guessing that this would need to be done using DESC resources at NERSC. Are you a DESC member by any chance?

Hi Jeff,

Thanks for following it up. I am a member of the DESC, but I have not used NERSC before. I will see if I can run the code in LSSTDESC.

Hi @Somayeh_K,

Just wanted to follow up on this to see if @jeffcarlin’s feedback was able to help resolve your issues with identifying the MLT objects. I’ve had some experience in working with this notebook, so I’m happy to look into it more if you need more help with it.

I marked Jeff’s response as the solution, because it did confirm that the lightcurves aren’t available in the parquet files, but could be via the DESC – and Somayeh did confirm that she would follow-up there.

In case anyone else finds this topic thread, note that the fact that no actual microlensing events were included in DC2 was confirmed in another thread: