O3b-Review.md

@@ -9,22 +9,17 @@ This is the home page for the review of the parallel version of the Bilby parame
 # Review Synopsis
 
 The review aims to establish that 
-1) Parallel bilby (pbilby) agrees with (serial) bilby using software injections 
-2) pbilby is consistent with LALInference on the GW events GW150914 and GW170817
-3) pbilby produces unbiased posteriors when phase marginalization is turned on and off 
-4) pbilby with both phase marg on and off is unbiased as shown by a pp test on 100 BBH injections, as is standard. 
+1) pbilby is consistent with the GWTC-1 LALInference and bilby runs on on the GW events GW150914, and others?...
+
+All review tests will be done with phase marginalization turned **off**, and time and distance marginalization turned **on**. 
 
 # Review statement
 
-> The review concludes that parallel bilby passes the proposed review tests above for BBHs on the git hash 359b3152e04573190dd4047402c390d662630b20. During the review, it was understood that the dynesty sampler used by parallel bilby can suffer from shredding (see slide 8 in [this presentation](https://dcc.ligo.org/DocDB/0166/G2000207/001/Updates%20on%20various%20bilby%20review%20issues%20for%20dynesty.pdf)). This bug can be clearly seen in traceplots and resolved by more conservative settings: the defaults are selected for GW150914, but in other higher-SNR events, it may be required to increase `nact` and/or `nlive`. We advise users to check traceplots for high SNR events durings runs.
-In the course of this review, we discovered difficulties reproducing data conditioning between bilby and lalinference. Understanding [differences](https://git.ligo.org/serguei.ossokine/pb_li_comparison/-/wikis/data_conditioning) in data between bilby and LI are crucial for any production runs using bilby, and fall within the scope of the review of core bilby modules, i.e., they fall outside the scope of the pbilby review. 
-The successful conclusion of this technical review is a necessary but not sufficient condition for the inclusion of pbilby results in the exceptional event papers and future collaboration results papers.
+
 
 # Update on review statement 11/05/20
 
-> A bug has been found in bilby (that also affects parallel bilby) when using the time marginalisation. This bug can have an effect on the sky position posteriors. The suggestion is to not use the time marginalisation at the moment. More investigations are being done.
 
-> A bug has been found in parallel bilby (at the moment bilby seems to be uneffected) that seems to affect the tilt_1 angle when sampling in chirp mass and mass ratio is done. It is suggested to run using the sampling in component masses. More investigations are being done about the bug.
 
 # Developer Team
 *  Greg Ashton