Remnant-fits.md

@@ -310,11 +310,11 @@ Maximum difference for af_evol_avg: 9.770286555355057e-07
 
 * see in-line discussion threads on !245
 * most other comments were resolved through the Mattermost chat. Some important topics/decisions:
-  * The last remaining differences against Nathan's reference implementation were traced down to only different redshift and q<->(m1,m2) treatments between PESummary and lalinferences.bayespputils, nothing to do with the NR fits or spin evolution.
-  * It was decided to only add a `_non_evolved` suffix if applicable, with evolved quantities being e.g. just `final_spin` without a suffic, to make life easier for external non-expert users of the released meta files.
+  * The last remaining differences against Nathan's reference implementation were traced down to only different redshift and q<->(m1,m2) treatments between PESummary and lalinferences.bayespputils, nothing to do with the NR fits or spin evolution. Very small differences in the masses (~1e-14) caused much larger differences (~1e-4) in the evolved spins.
+  * It was decided to only add a `_non_evolved` suffix if applicable, with evolved quantities being, e.g., just `final_spin` without a suffix, to make life easier for external non-expert users of the released meta files.
   * For the same reason, the evolved angles are not stored in the final output file.
   * Reference frequency etc are stored in the metadata only, not repeated for each sample.
-  * The fits used are also added to the metadata, but the approximant used for spin evolution is not (as this cannot currently be changed by the user, so is uniquely reconstructable from the PESummary version number).
+  * The fits used are also added to the metadata, but the PN approximant used for spin evolution is not (as this cannot currently be changed by the user, so is uniquely reconstructable from the PESummary version number).
 
 ## sign-off