Telecon Notes Oct. 18 2018

Attendance: Kris, Graca, Soumen
Regrets:

Notes: Shaul

• Status Updates

• Most important outcome: there will be another telecon on Monday at 10:15 am Central Time to discuss new results from Soumen. Information will be circulated.
• Soumen:
  • Puzzle with TT, EE maps and spectra not yet resolved. Didn't have much time to try to debug, but will continue to search for the issue. For the time being, let's concentrate on BB alone
  • He is expecting to get results from running 10 realizations of all models, all delensing values by tonight. Perhaps 10 more by Monday.
  • There will be a telecon to discuss these results by Monday (see above).
  • The BB map will be available for Graca to analyze.
• Graca:
  • Graca ran Belen's results on model 90. Check out the recent posting in the logbook. Generally the results look OK, but the BB r likelihood peaks at zero. Why? Didn't Belen use r=0.003 for the input? Also, the most likely BB power spectrum model shows lower power than the data for lensing. Not clear why. Are the inputs correct? Graca needs to check with Belen and Clem what is used for inputs.
  • SH points out that after this is sorted, Belen has already made available maps for models 91 and 92. Could Graca please run these after the puzzles are sorted out? Answer: Graca wasn't aware that other maps are available. Yes, she will run
  • Graca reports that Aurora, the JPL machine on which she was running the parameter estimation has been reserved next week for another project. She will try to install her code on Halo. She can't run this at NERSC because NERSC has decommissioned the Intel-13 compiler she needs.
  • Shaul suggests checking also at the UMN superpercomputing center. Graca and Shaul will follow up.