| ... | ... | @@ -4,12 +4,24 @@ |
|
|
|
|
|
|
|
- **Stefano** - remove patch to print likelihood terms from precessing build since it seems to break the dag; let Shio know about this issue and use her newer version if it fixes this issue; figure out how to do this without modifying the build (ie move modified libraries to run directory)
|
|
|
|
- **Pratyusava** - check behavior of '--spin-aligned' option. May need bugfix
|
|
|
|
|
|
|
|
- **Stefano** - add instructions to precessing wiki on how to run with Gaussian white noise
|
|
|
|
|
|
|
|
## Round-table Updates
|
|
|
|
|
|
|
|
Stefano
|
|
|
|
- Support for IMRPhenomXPHM should now be added to precessing branch
|
|
|
|
- Started some runs with Gaussian white noise; things haven’t finished but same error with 16 failed jobs is probably showing up
|
|
|
|
- Will make wiki in precessing section about how to do these sorts of runs
|
|
|
|
- Shows chisq-snr plots to support that IMRPhenomXPHM has better behavior;
|
|
|
|
- For high precessing spin, IMRPhenomPv2 has weird behavior; Sarah - how do we understand this?
|
|
|
|
- Khun Sang – most of precession should depend on inclination not theta angle
|
|
|
|
- Stefano – but these templates have inclination set to zero so they are always set to face-on; if you take highly precessing waveforms – a lot of in-plane spin; Pv2 gives some weird behavior
|
|
|
|
- Amit – theta is angle between theta_JM; high value of theta_JM gives more precessing
|
|
|
|
- Stefano – not using theta_JM but these variables can translate to that
|
|
|
|
- Shows distribution of likelihood ratio values for missed injections; appears to have no clear distinction between missed and found and no terms that are obviously bad
|
|
|
|
- Sarah - Can we see missed/found lnL and FARs plotted? These should show clear distinction between two populations. And there should be a 1-to-1 mapping between the two. If not, we need to understand why not. If it works, see if you can calculate the lnL value reported by the search by using the terms printed out from the script.
|
|
|
|
- logL in coinc_event tables
|
|
|
|
- What are fars that get printed out with Shio's script?
|
|
|
|
|
|
|
|
Pratyusava - issues with some runs
|
|
|
|
- having some problems to generate injections
|
| ... | ... | @@ -24,29 +36,11 @@ Pratyusava - issues with some runs |
|
|
|
- There seems to be issue with –spin-aligned option showing up in configuration. Pratyusava will check this.
|
|
|
|
|
|
|
|
Khun Sang
|
|
|
|
|
|
|
|
– data for precessing injections; data from pycbc run
|
|
|
|
- data for precessing injections; data from pycbc run
|
|
|
|
- PyCBC injection files at CIT: /home/pycbc.offline/O3/production/O3B_CATALOG/ALL_TRIGGER_FILES/FOCUSED_BBH/HDFINJFIND
|
|
|
|
- These are hdf5 files. Simple script to read those files: CIT:/home/pycbc.offline/O3/production/O3B_CATALOG/INJECTION_SUMMARY/HYPERBANK/list_results_csv.py
|
|
|
|
- Gstlal injection files at CIT: /home/pycbc.offline/O3/production/O3B_CATALOG/ALL_TRIGGER_FILES/FOCUSED_BBH/HDFINJFIND
|
|
|
|
- Are these the same as what is in Zenodo? Zenodo is missing snr and chisq information
|
|
|
|
|
|
|
|
Amanda – trouble generating summary pages; failures in calc_rank_pdf jobs: "index is out of range"
|
|
|
|
- Might be related to changes related to print likelihood terms
|
|
|
|
|
|
|
|
Stefano – running with white noise, things haven’t finished but same error is probably showing up; simple to run with white noise
|
|
|
|
- Make wiki in precessing code about how to do this
|
|
|
|
|
|
|
|
- High precessing spin: IMRPhenomPv2 has weird behavior;
|
|
|
|
|
|
|
|
- Khun Sang – most of precession should depend on inclination not theta angle
|
|
|
|
|
|
|
|
- Stefano – but these templates have inclination set to zero so they are always set to face-on; if you take highly precessing waveforms – a lot of in-plane spin; Pv2 gives some weird behavior
|
|
|
|
|
|
|
|
- Amit – theta is angle between theta_JM; high value of theta_JM gives more precessing
|
|
|
|
|
|
|
|
- Stefano – not using theta_JM but these variables can translate to that
|
|
|
|
|
|
|
|
Get logP in coinc_event tables
|
|
|
|
What are fars?
|
|
|
|
SNRs joins might not be same |
|
|
|
- Might be related to changes related to print likelihood terms. Stefano will fix and Amanda can try again |
