Added a maximum threshold of 1000 km/s for Disberg & Mandel (2025) LOGNORMAL kicks, matching paper#1405
Conversation
…GNORMAL kicks, matching paper
There was a problem hiding this comment.
Approved with a couple of questions:
(i) on line 3911 you have the comment
"makes it possible to adjust if p_Rand is too low, to avoid getting stuck"
but on line 3914 you reduce rand. Does the comment correctly/completely capture what you mean?
(ii) should the threshold value of 1000.0 be a constant defined in constants.h (named something like DISBERG_MANDEL_MAX_KICK)? Then the 1001.0 in line 3910 can just be (e.g.) DISBERG_MANDEL_MAX_KICK + 1.0.
EDIT: also, rand = 0.99 * rand; can just be rand *= 0.99; (more succinct, and not too obscure...)
|
I ran COMPAS v03.20.06 with ./COMPAS -n 100000 --mode SSE --kick-magnitude-distribution LOGNORMAL. From the results I selected the NS-forming SNe with SN_Type==1 and in the attached figure I plot a CDF of the values of "Applied_Kick_Magnitude" (which is exactly equal to "Drawn_Kick_Magnitude"). For v<900 km/s it exactly follows our log-normal distribution, if it is not normalized between 0 and 1000 km/s. Between 900 and 1000 km/s, there is a clear feature in the COMPAS results that makes them deviate from the log-normal and make the total distribution normalized for v<1000 km/s. Could it be that the threshold of 1000 km/s is currently causing all kicks with v>1000 km/s to be projected on the region between 900 and 1000 km/s? That is, in the PDF there is a clear peak between at v>900 km/s, so I believe something is not going right with the normalization. (This feature is, e.g., not present in the figure @SimonStevenson made in #1404.) |
|
Thank you, @pauldisberg and @jeffriley . I was also taking advantage of this PR to do a bit of work on the Matlab post-processing, which doesn't need to be reviewed. Done with that for the day, so merging in. |


No description provided.