Request for official definition (or list) of fₚₑₐₖ and reproducible method — our results don’t match 150/188/85 Hz

[sahsouh]

Hi all,

I’m trying to compute the “peak frequency” fₚₑₐₖ for a few BBH events using the public time-series waveforms (CSV) for GW150914, GW151012, and GW170729, but my numbers don’t match the commonly quoted values (≈150, 188, and 85 Hz, respectively). I’d really appreciate either (a) an official list of fₚₑₐₖ for these events or (b) a precise, reproducible method so I can compute them the same way the catalogs/figures do.

What I did (single rule, applied identically to all events)

• Goal/definition I used: frequency at the peak of the amplitude (“merger”) of the time-domain strain.

• Procedure (no event-specific tuning):

  1. Remove DC;

  2. Find the amplitude peak time t\* via the Hilbert envelope;

  3. Extract a 0.4 s window with the peak positioned at 55% of that window;

  4. Compute the instantaneous frequency at t\*​ from the Hilbert phase derivative (Savitzky–Golay smoothing ~15 ms).
     → Call that value fₚₑₐₖ.

What I get vs. what I expect

• GW150914: measured ~186 Hz vs. expected ~150 Hz

• GW151012: measured ~303 Hz vs. expected ~188 Hz

• GW170729: measured ~118 Hz vs. expected ~85 Hz

I’ve attached a quick bar plot for clarity:
Download figure: f_peak_comparison.png

What I’m missing / questions

It seems the discrepancy is about conventions, not data quality. Could you please clarify the catalog definition of fₚₑₐₖ?

1. Exact definition: Is fₚₑₐₖ

   • the frequency at the maximum of |h| (strain) in the detector frame,

   • or the frequency of the (2,2) mode at the maximum of |h₂₂| from NR/IMR models,

   • or the frequency at the peak of the GW luminosity Lpeak​,

   • or something else?

2. Preprocessing: Is the strain whitened before measuring fₚₑₐₖ? If yes, which PSD and bandpass are used (e.g., fixed 20–512 Hz)? Any window length around the peak that’s standard?

3. Frame: Are catalog fₚₑₐₖ values quoted in the detector frame or source frame? (For high-z events like GW170729 the difference matters.)

4. Official list: If there’s already a vetted list of fₚₑₐₖ for O1/O2/O3 events (with the above conventions), could you point me to it?

I’m happy to adopt the official pipeline/definition; my goal is simply to produce fₚₑₐₖ values that match the catalog conventions across events without event-by-event tuning.

Many thanks!

[jonah]

@sahsouh Thanks for the great question! I’m sorry, I don’t have a quick answer for you. Can you point me to an example of where you found f_{peak} published? That would help me understand the context for this question.