... | ... | @@ -155,20 +155,34 @@ distance-marginalization = True |
|
|
* To run `bilby`, run `run_bilby.py`
|
|
|
* To run `pbilby`, we first run `parallel_bilby_generation parallel.ini` and then submit the job in the usual way
|
|
|
|
|
|
For the comparions below, we run both parallel bilby and bilby 4 times and combine the results using `bilby_result -r *result.json --merge`.
|
|
|
|
|
|
## Timings
|
|
|
|
|
|
All timings measured on `sstar` which uses 2 Intel Xeon E5-2660 (8 core) processors.
|
|
|
|
|
|
| Sampler | Average wall time (over 4 runs) |
|
|
|
| ------ | ------------------------------- |
|
|
|
| Parallel bilby 50 nodes with 16 cores per node | 11m |
|
|
|
| Serial bilby | 4126m = 68.7hr |
|
|
|
|
|
|
|
|
|
## Summary Pages
|
|
|
* [PESummary page comparison](https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/parallel_bilby/bilby_comparison/bilby_comparison_10/comparison/home.html)
|
|
|
* [PESummary page: all plots in one](https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/parallel_bilby/bilby_comparison/bilby_comparison_10/comparison/plots)
|
|
|
* [PESummary page comparison](https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/parallel_bilby/bilby_comparison/bilby_comparison_11/comparison/home.html)
|
|
|
* [PESummary page: all plots in one](https://ldas-jobs.ligo.caltech.edu/~gregory.ashton/parallel_bilby/bilby_comparison/bilby_comparison_11/comparison/plots)
|
|
|
|
|
|
## Selected comparisons
|
|
|
|
|
|
### Masses
|
|
|

|
|
|

|
|
|
|
|
|
### Spins
|
|
|

|
|
|

|
|
|
|
|
|
### Others
|
|
|

|
|
|

|
|
|
|
|
|
### SNRs
|
|
|
 |
|
|
\ No newline at end of file |
|
|

|
|
|
|
|
|
Note: the SNRs in bilby are stored as complex numbers. The optimal SNR has no complex part, but to ensure that both codes are doing the same thing I plot out both the real and complex parts. |
|
|
\ No newline at end of file |