Getting started tutorial part 2 (and 4): qgraph problems

Support
#1

Hello,

I have been working my way through the getting started tutorial. I had some trouble with setting up the Butler data repository in part 1, which I eventually passed using the helpful instructions here.

For step 2, I was initially using the tutorial located at Getting started tutorial part 2: calibrating single frames with Pipeline Tasks — LSST Science Pipelines which I couldn’t get through. I switched to the /v/weekly version of the tutorial, but was still unable to pass the pipetask run step, as I don’t appear to have a DRP-RC2_subset.yaml file. In that folder ($DRP_PIPE_DIR/pipelines/HSC/) I located DRP-RC2.yaml, so decided to try using this instead. This seemed to work as I was able to run the pipetask run command and then display some images in DS9 (following the tutorial part 3), but I couldn’t get pipetask qgraph to work with the same tweak. It appears this might be causing the first step of tutorial part 4 to fail, as I am receiving an error message “RuntimeError: QuantumGraph is empty.”

Now I’m stumped. Could anyone offer some advice on how to fix this issue?

#2

I am not this tutorial’s author — hopefully they can jump in with a definitive answer. But your RuntimeError: QuantumGraph is empty. message reminds me of something I encountered in early 2022 when running the getting started tutorial myself. In case it’s of interest, here’s the link to the relevant thread, starting with the comment that has this same error message:

#3

Thanks for your response, Aaron. It helped me make some progress.

Reading through that thread, I realized that I had a version mismatch. The rc2_subset I initially cloned from github was on the main branch, which has no pipelines folder. I started over with the v24_0_0 of the tutorial, got the correct rc2_subset, and repeated parts 1, 2, and 3. I attempted part 4 and failed on the FGCM step. When running

pipetask run -b $RC2_SUBSET_DIR/SMALL_HSC/butler.yaml \
             -p $RC2_SUBSET_DIR/pipelines/DRP.yaml#fgcm \
             -i u/$USER/single_frame \
             -o u/$USER/fgcm \
             --register-dataset-types

I get a bit further than before (no quantumgraph is empty error) but still run into a problem. One quanta executes successfully, but then I get RuntimeError: FATAL: All observations in i band have been cut with TOO_FEW_STARS

Attached is the full output of that step.
fgcm.txt (12.6 KB)

#4

It seems that you found the solution to your problem, which is that you must use the same version for rc2_subset as for the rest of the pipelines. Unfortunately, older versions of the docs don’t specify this, whereas newer versions don’t seem to specify the correct version - we’re looking into this.

As for the FGCM failure, @erykoff might have an idea whether that failure was expected in v24 or not. Regardless, though, I suggest trying to go through the tutorial with a more recent weekly.

#5

@jeffcarlin any advice regarding this topic? Thanks…