I read the article. samshap is 100% right and you are 100% wrong.
[EDITED to add:] “What is asserted without evidence can be dismissed without evidence”—but I might as well give some justification for my claim. Here is what the article says:
First I begin by cross-correlating the Hanford and Livingston data, after whitening and band-passing, in a very narrow 0.02s window around GW150914. This produces the following:
(followed by the graph you provide here). Note two things.
First: “I begin by cross-correlating the Hanford and Livingston data”—just as samshap says.
Second: “in a very narrow 0.02s window”. That’s about 1⁄10 of the time period represented by the main plots, which go from 0.25s to 0.45s “relative to September 14, 2015 at 09:50:45 UTC” (not that we can tell from your presentation, because you clipped off the bottom part of the figure which includes the time axes). So this could not possibly be an alternative to the other plots; the horizontal axes aren’t in any way compatible.
The context for this is that the (LIGO-skeptical) Cresswell et al paper is looking at the time lags between LIGO observations, and claiming to cast doubt on the idea that seeing two very similar signals at the two detectors at a certain time-lag is evidence of anything. So, in particular, Cresswell et al try to show that you can get the same 7ms lag by looking at other things without the actual signal in it. (One of the things they look at is the residual noise from the LIGO data, after subtracting off the black-hole-merger model. This is why it’s relevant that the actual best-fit model is better than the “illustrative” one—because if you subtract off a crude model, what remains will have some real signal in it, so it’s unsurprising if it shows some of the same temporal correlations as the actual signal does.) So now Ian Harry shows the cross-correlation graph for the LIGO data before subtracting off the fitted model, and after subtracting the (best) fitted model. The graph you reproduce here is the cross-correlation before subtracting the model; the next one (not reproduced here) is the cross-correlation after subtracting the model, which shows no 7ms spike.
Note that the context makes excellent sense of having a cross-correlation graph at this point in the article, and would make no sense at all of having a raw-LIGO-observation-data graph instead.
I read the article. samshap is 100% right and you are 100% wrong.
[EDITED to add:] “What is asserted without evidence can be dismissed without evidence”—but I might as well give some justification for my claim. Here is what the article says:
(followed by the graph you provide here). Note two things.
First: “I begin by cross-correlating the Hanford and Livingston data”—just as samshap says.
Second: “in a very narrow 0.02s window”. That’s about 1⁄10 of the time period represented by the main plots, which go from 0.25s to 0.45s “relative to September 14, 2015 at 09:50:45 UTC” (not that we can tell from your presentation, because you clipped off the bottom part of the figure which includes the time axes). So this could not possibly be an alternative to the other plots; the horizontal axes aren’t in any way compatible.
The context for this is that the (LIGO-skeptical) Cresswell et al paper is looking at the time lags between LIGO observations, and claiming to cast doubt on the idea that seeing two very similar signals at the two detectors at a certain time-lag is evidence of anything. So, in particular, Cresswell et al try to show that you can get the same 7ms lag by looking at other things without the actual signal in it. (One of the things they look at is the residual noise from the LIGO data, after subtracting off the black-hole-merger model. This is why it’s relevant that the actual best-fit model is better than the “illustrative” one—because if you subtract off a crude model, what remains will have some real signal in it, so it’s unsurprising if it shows some of the same temporal correlations as the actual signal does.) So now Ian Harry shows the cross-correlation graph for the LIGO data before subtracting off the fitted model, and after subtracting the (best) fitted model. The graph you reproduce here is the cross-correlation before subtracting the model; the next one (not reproduced here) is the cross-correlation after subtracting the model, which shows no 7ms spike.
Note that the context makes excellent sense of having a cross-correlation graph at this point in the article, and would make no sense at all of having a raw-LIGO-observation-data graph instead.