Research Journal (Summer 2020)

Thursday, June 25

  • Continued working on 15bn analysis, working on the first time bin (which hadn't converged yet) using jupyter notebook
  • Met with Prof Mukherjee to discuss longer term plan - after finishing 15bn analysis, could move on to analysis of the new SLSNe source. After that we will need to decide whether to move on to fermi analyses of different type of sources (like PWN) or whether to stay with the SLSNe project in a different way
  • Read journal club article
  • Attended journal club

Wednesday, June 24

  • Send Deivid the 17egm comparison powerpoint (also attached that presentation on this page), we will spend some time soon going through all of the results in detail
  • Finished the initial run through of 15bn with tehanu fermipy (all bins) -- not all of the bins converged, so will need to work on playing with those models more to make them converge, but they ran successfully and gave results, which is great
  • Met with Deivid, discussed the 17egm comparisons (how/what exactly to compare considering that my index was fixed and his was free), and talked about the next steps for 15bn (specifically discussed using jupyter notebook to allow a more interactive way of changing the model in real time by fixing/freeing/deleting sources
  • Group meeting - discussion of whether to rerun 17egm with free index (decision to hold off for the moment, maybe return to it after I finish up with 15bn), and more discussion of which output products are best to compare in this situation
  • APS webinar - Making Physics Inclusive and Equitable

Tuesday, June 23

  • Solved the Invalid DISPLAY variable error -- switched to using the -Y flag for ssh instead of the -X flag
  • Successfully got the 15bn analysis running
  • Finished organizing 2017egm files, put the results in a powerpoint with Deivid's results for easy comparison -- as I was looking at the results, I noticed that the model that I used was a very constrained one (had all index parameters fixed, even for 17egm itself). The fits did converge, but it was not the same model that Deivid used, and probably did not get as accurate results.
  • Attended Summer Colloquium -- Dr. Zsuzsanna Marka

Monday, June 22

  • Finished getting fermipy set up, made config files, made time cuts to remove time periods where the sun is in the area before starting the analysis
  • Tried to run analysis for one time bin of 15bn on fermipy, ran into a new error (RuntimeError: Invalid DISPLAY variable), but did successfully get past the point where fermitools was failing before
  • Group meeting
  • Working on getting 2017egm files organized

Thursday, June 18

  • Finished reading journal club article
  • Started getting fermipy set up, following Ari's slides as well as looking at general fermipy information online
  • Attended Daniel Savin's colloquium
  • Attended journal club - in addition to discussing the article, made up rotation plan for the summer (I will be leading journal club on July 9)

Wednesday, June 17

  • Tried running analysis step by step rather than using my script (on tehanu), in case the script was somehow the issue, but it still didn't work
  • Discovered that when I transfered my 2017egm files between computers they got somewhat mixed up and out of order, so I will need to spend some time getting them sorted out before I can compare results directly to Deivid's (didn't have time today)
  • Met with Deivid
  • Started reading the article for Thursday's journal club
  • Group meeting - general consensus that I should switch over to using fermipy because it doesn't look like we will be able to get fermitools working again soon

Tuesday, June 16

  • Tried to run PG1553 on tehanu without using files from my computer (accessing data and creating files completely through tehanu) - ran into a floating point error at gtsrcmaps step
  • So, currently have two issues: the out of range error at likelihood step on all machines regardless of data/model, and this new floating point error at gtsrcmaps on tehanu (it works fine on my computer)
  • Spent some time looking around online trying to see if anyone has had these issues - I did find someone wrote in on the fermi github about a likelihood out of bounds error, but the issue was closed without there being any comments, so I'm not sure what to make of that
  • 1pm - Attended Prof Mukherjee's colloquium
  • 4pm - Attended Prof Mike Nowak's Washington University talk on high energy astrophysics analysis softwares
Monday, June 15
  • Started working on getting my SN2017egm results from last summer organized so I can start comparing the results with Deivid's results
  • Tried running PG1553 analysis on tehanu, had to deal with a few minor issues to start running
  • a couple of minor issues - had to figure out all the right paths for the files and scripts. bdlikeSED uses pyfits, which doesn't seem to exist anymore, so had to change it to using
  • Attended group meeting
  • The run failed at the same place and with the same error as it did when running on my machine. Will look into this more tomorrow.

Friday, June 12

  • Checked the fits files involved in the analysis, there doesn't seem to be any issue of corrupted files
  • Worked on getting set up with tehanu and Ari's fermipy setup
  • First thing I want to try is moving the PG1553 files onto tehanu and running their analysis there (because I am more confident with that model than with the one for SN2015bn, so I will test it first before moving on to 15bn)
  • Ended up having to move the bdlikeSED and make4FGLxml scripts over from my computer, I don't know how to access them within the tehanu setup (since they aren't part of the standard fermitools)

Thursday, June 11

  • Met with Deivid - discussed current status of fermitools troubleshooting (still haven't fixed it), as well as ideas for moving forward on that and other things
  • Goal to compare my fermi results for 2017egm from last summer with Deivid's results on the VERITAS wiki - first step will be to look through my own results to make sure they are all in order and complete
  • Last idea for troubleshooting fermitools - look at the fits files using astropy to make sure they are not corrupted
  • However, since it seems unlikely that we will solve the fermitools issue soon, making plans to switch to using either Qi or Ari's setup on tehanu
  • Had an issue with getting onto tehanu, but Bill Seligman helped me get it fixed, so it should be working now

Wednesday, June 10

  • Particles for Justice strike

Tuesday, June 9

  • Continued fermitools troubleshooting - tried running without 15bn in the model, double checked that I'm using most up to date diffuse models, tried using different optimizers
  • None of these things solved the issue, and I am currently out of ideas for what to try next
  • Finished reading the X-ray paper

Monday, June 8

  • Continued troubleshooting: tried reinstalling fermitools and anaconda, didn't help with the error. Trying to look online to see if anyone has had similar issues with fermitools, so far no luck.
  • Colin sent out a doodle poll for times for journal club
  • Group meeting - discussion of Wednesday's strike for social justice and the summer colloquium series (sign up on link)
  • Some fermi suggestions from group meeting: try running analysis with SN2015bn out of the model, make sure using the most up to date diffuse models, try using different optimizers.
  • Qi made a setup of fermitools on tehanu last year, if it is still up and I can't get my setup working maybe I can try using my files in that setup
  • Next Wednesday undergrads will give a short presentation on background material after reading papers
Thursday, June 4 - Friday, June 5
  • Continued working on troubleshooting fermi analysis, spoke more about it with Deivid via slack
  • I've determined that the issue is not specific to the particular model file I was using, or to 2015bn as a source (I tried different models, and did a test analysis with a different source, which also failed)
  • The next step will be trying to completely reinstall fermitools and see if that helps

Wednesday, June 3

  • Continued to work on troubleshooting fermi analysis

  • Met with Deivid to discuss background theory of SLSNe and the magnetar model, as well as to work on troubleshooting the error

  • We came up with some ideas to work on, including looking through the xml file to make sure there are no obvious issues, trying to run the analysis for other time bins, and changing the model. Also discussed strategies for how to choose to fix/free sources in the model in general.

  • Group meeting

Tuesday, June 2

  • Tried to work on fermi analysis of SN2015bn, ran into error at the actual likelihood step "Attempt to set the value outside of existing bounds. Value nan is not between 1e-05 and 1000"

  • (was successful at preparing the preliminary files with all of the time cuts to remove contamination from the sun)

  • spent the rest of the day troubleshooting

    • I determined that I had updated Fermitools recently, so I tried undoing the update, but it didnít make a difference.

    • Tried using different model files, including one which had worked in an earlier analysis, but continued to get the same error

Monday, June 1

  • Various orientation meetings, discussed plans for the summer. Primary goal is to finish the SLSNe analyses and get more familiar with the theory.

  • Attended Ari's fermipy tutorial

  • started reading the X-ray SN2015bn paper

-- Pazit Rabinowitz - 2020-06-03


Edit | Attach | Watch | Print version | History: r45 | r16 < r15 < r14 < r13 | Backlinks | Raw View | Raw edit | More topic actions...
Topic revision: r14 - 2020-06-26 - PazitRabinowitz
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback