Optimizing black hole merger waveform calculations (Bluesky thread)

ETA: this gathers up the bluesky thread readably https://skyview.social/?url=https%3A%2F%2Fbsky.app%2Fprofile... although you'll probably prefer "unroll" over "tree" (but "tree" seems to capture everything including a handful of replies).

Cross-posted to x/twitter: <https://twitter.com/duetosymmetry/status/1769913120099545488>

(ETA: there are 38 (thirty eight) entries in the thread. Logged in on x/twitter I see them all, but bsky (logged in and not) sometimes doesn't show them all in one scroll-through. :/ )

tl;dr (it's a very technical explainer thread)

[preliminary] this is physics, it is important to match GW (and optical) detections of mergers, not to model "spherical cows"

[a] post-Newtonian (PN) methods are cheap to calculate for black hole binaries when they are far apart (and other methods are good for extreme mass ratios) but in the last moments before merger we need numerical relativity (NR). NR is expensive even on supercomputers.

[b] PN and NR are different enough frameworks that knitting together the early inspiral PN and the late pre-merger NR requires some overlap and iteration.

[c] results from a couple dozen simulations give interesting technical insights

[d] orbital eccentricity (and ellipticity) make everything harder, especially for PN, because the plane of the orbit and the orientation of the black hole spins against the orbital plane all precess (and differently from each other)

Finally, if you can't see the thread on either platform, the prerpint is at https://arxiv.org/abs/2403.10278

(I did double-check and can read the bsky thread via the Tor browser (obvs. not logged in). x/twitter is ... uncooperative when not logged in. Finally, mastodon threading is ... confusing.)