Merged Discussion of femr's video data analysis

I've mentioned implications of the NIST errors several times. Did you miss them ?

Continnued insistence from some members that the NIST values are correct, and femr2 is wrong... keeps NIST "on the slab".


Correct. Build an extremely intricate model, then slam the aircraft model into the tower model in the wrong direction. Doh ! :rolleyes:

To be fair you haven't even shown that NIST made an error. The "NIST" aircraft in your comparison is at completely the wrong angle and the quoted angles in your draft are in the wrong frame of reference and therefore wrong. My assessment shows the NIST values to be correct within their margin for error.

Your basic assertion that NIST was 9-10° out with their impact orientation needs serious reassessment.
 
Sigh...
To be fair you haven't even shown that NIST made an error.
It's rather obvious that they have, and the various rotoscope views all agree with my assessment.

That you are not "convinced" is not a problem.

The "NIST" aircraft in your comparison is at completely the wrong angle
Until I relocate the reference mentioned earlier, I'm forced to say...yeah, the NIST aircraft in my (three year old) rotoscope is out a couple of degrees.

Annoying, but not a huge issue.

the quoted angles in your draft are in the wrong frame of reference
NIST use various frames of reference, including that of the aircraft. Using the building as the frame of reference for an aircraft yaw, pitch and roll is "rather odd" (to say the least), which is why NIST vary the frame of reference within the report I imagine. Different "authors" having different viewpoints.

Again, until I find the reference mentioned earlier, I'm forced to say... yeah, the NIST aircraft in my rotoscope views uses the aircraft as the frame of reference, not the building.

and therefore wrong
Puts the NIST aircraft in the wrong orientation, but not mine.

I've already restated my values in the alternate frame of reference.

My assessment shows the NIST values to be correct within their margin for error.
It really doesn't. If you want to "contest" that statement, please reference the NIST error band for their 13 degree lateral value, then determine the "actual" lateral angle. As always, you should get to my orientation as you dig into the technicalities further.

The issues raised put the NIST aircraft "closer" to actual, but still pretty far off.

Your basic assertion that NIST was 9-10° out with their impact orientation needs serious reassessment.
We've already "been there, done that". It chopped a couple of degrees off the NIST lateral error. Plenty remaining.

You appear to be "forgetting" updated values...
 
From another thread:

Savitzky-Golay, Poly (10) and Poly (50) are three different methods of smoothing that produce somewhat different results, as shown in the graph.

"NIST" means the derivative of NIST's nonlinear fit curve given in the WTC7 report.

ETA: In short, the graph shows that the acceleration does not increase instantly to g.

That's mostly an artifact of the smoothing. With better analysis, it's easy to see the onset of acceleration was considerably quicker than you'd conclude from that graph.
The graph in question is this one (famous in this thread and others):

http://femr2.ucoz.com/_ph/7/513801604.png

However, since I could not reproduce it, I'll be using the black one in this graph instead, which is the one I generated (it's compared to femr2's for reference, which is the red line):

http://www.formauri.es/personal/pgimeno/xfiles/11-s/femr2-qtiplot-comparison.png

Here it is without the comparison overlaid:

http://www.formauri.es/personal/pgimeno/xfiles/11-s/SG1-SG2-Vertical-Acc.png

Raw numeric data available on request.

This graph shows a general descent from zero acceleration at second ~11.9 to about -38.16 ft/s² at second ~12.8, although that line has a few local "bumps".

It corresponds to the following fragment of the raw position data:

RawPositionDataZoomed01.png


Looking at that graph, there's a quite visible jerk (an abrupt change in the visible mean slope) at about second 12.55.

In fact, a quadratic regression over that whole interval does a poor job:

RawPositionDataZoomed01WithQuadraticFit.png


Code:
Polynomial Fit of dataset: RawPositionData_2, using function: a0+a1*x+a2*x^2
Weighting Method: No weighting
From x = 1.1911911911910e+01 to x = 1.2796129462800e+01
a0 = -1.0905789568220e+03 +/- 9.5651463026293e+01
a1 = 1.7937822562044e+02 +/- 1.5489877275163e+01
a2 = -7.3811148105196e+00 +/- 6.2688869526111e-01
--------------------------------------------------------------------------
Chi^2/doF = 7.7529164984578e-02
R^2 = 0.9166610632945
Adjusted R^2 = 0.9116607270922
RMSE (Root Mean Squared Error) = 0.2784405950729
RSS (Residual Sum of Squares) = 3.953987414213

The mean acceleration for that period would be twice a2, or -14.7±1.3 ft/s².

So I've done two fits, one for the part before 12.55 and one for the part after 12.55:

Zoomed01DoubleQuadraticFit.png


Red line fit data:
Code:
Polynomial Fit of dataset: RawPositionData_2, using function: a0+a1*x+a2*x^2
Weighting Method: No weighting
From x = 1.1911911911910e+01 to x = 1.2545879212550e+01
a0 = -1.6806123247285e+02 +/- 9.2689508776037e+01
a1 = 2.8316558594865e+01 +/- 1.5161600794194e+01
a2 = -1.1985547453467e+00 +/- 6.1989424957249e-01
--------------------------------------------------------------------------
Chi^2/doF = 1.4872690196531e-02
R^2 = 0.7266837376128
Adjusted R^2 = 0.7032566294082
RMSE (Root Mean Squared Error) = 0.1219536395379
RSS (Residual Sum of Squares) = 0.5354168470752

Blue line fit data:
Code:
Polynomial Fit of dataset: RawPositionData_2, using function: a0+a1*x+a2*x^2
Weighting Method: No weighting
From x = 1.2545879212550e+01 to x = 1.2796129462800e+01
a0 = -3.2997572822317e+03 +/- 1.7413725888028e+03
a1 = 5.3187442084546e+02 +/- 2.7486656510457e+02
a2 = -2.1439756710978e+01 +/- 1.0846242220708e+01
--------------------------------------------------------------------------
Chi^2/doF = 5.2057084600355e-02
R^2 = 0.9490840513056
Adjusted R^2 = 0.936355064132
RMSE (Root Mean Squared Error) = 0.2281602169536
RSS (Residual Sum of Squares) = 0.6767420998044

That does a much better job at matching the position data. Now, the average acceleration for the first period is twice a2, or about -2.4±1.3 ft/s², and for the second period, it's about -42.9±21.7 ft/s². Arguably, the second part is too short to take that average acceleration as accurate, and indeed just by rejecting the first two samples of that interval, the fit is quite different (-64.2±31.8 ft/s²).

I wonder if that jerk is what you meant by "the onset of acceleration was considerably quicker than you'd conclude from that graph".

I'd love to compare that graph with the much more stable Cam #3 trace data, but I don't have a correspondence of the X axes of both yet.
 
I join carlitos in applauding.

Everything I'm going to say in this post is based upon calculations from a year and a half ago, when I abandoned this subforum out of disgust. Because those calculations shed some light (or darkness) on the other thread pgimeno mentioned, I'll finish my long-interrupted series on Savitzky-Golay differentiation and smoothing. I don't have time for much beyond that, and I'm dealing with a hand injury that makes it hard to use a keyboard.

The graph in question is this one (famous in this thread and others):

http://femr2.ucoz.com/_ph/7/513801604.png

However, since I could not reproduce it,
To extract numbers from that graph, I had to write a little computer program that digitized the red curve.

I wonder if that jerk is what you meant by "the onset of acceleration was considerably quicker than you'd conclude from that graph".
No, I was mainly remarking upon the fact that wide-window Savitzky-Golay differentiation is inherently incapable of revealing the abruptness of truly sudden changes in acceleration. The change in acceleration will be smeared out over the window size.

True SG differentiation/smoothing has that problem, and whatever femr2 did made the problem even worse.

Eighteen months ago, it took me only a few hours to construct a stepwise (hence maximally abrupt) acceleration curve that was more consistent with femr2's position data than is femr2's so-called "Savitzky-Golay" curve. That stepwise curve went from 0 to 32 ft/s^2 at 12.4s and to 40 at 12.7s. A straightforward error analysis says those transitions should have been placed a bit earlier, so the second step of my stepwise curve probably corresponds to what you found at 12.55s.

But the fact that my stepwise curve explains the position data at least as well as femr2's "Savitzky-Golay" curve means it is a grave mistake to say femr2's "SG" curve is evidence against sudden acceleration. I'll say more about that later today.
 
Savitzky-Golay smoothing, part 3

Eighteen months ago, I posted the first two parts of a planned series on Savitzky-Golay smoothing.

Savitzky-Golay smoothing, part 1 pointed out that femr2's original "Savitzky-Golay" acceleration curve was not obtained using true Savitzky-Golay differentiation.

Savitzky-Golay smoothing, part 2 described Savitzky-Golay differentiation and smoothing in some detail, explaining the historical importance of some details that don't matter so much today.

This long-delayed third part of the series will explore two technical points in greater detail. From part 2:

For many applications, the most questionable choice is the decision to give equal weight to the squared error at each point. Most methods give less weight to the error at points farther away from the center of the window.

...snip...

this feature of Savitzky-Golay smoothing may give too much influence to outliers or noise at the extremes of a window. It also creates lag when wide windows are used to smooth a time series for position that may have resulted from sudden changes in acceleration.

That could be of concern when analyzing video of building collapses. To find out whether it's a problem, we'd have to understand the purposes of the analysis.
It's been hard to identify any possible purpose for the various acceleration curves produced by femr2's video analysis, but we recently saw several of those curves being used to argue against David Chandler's claim that WTC7 "transitioned from rest to freefall essentially instantly". The person who did that soon regretted his introduction of those curves into that debate, but this incident shows why we can't judge the usefulness of an analysis without considering its intended purpose(s).

As shown below, none of the cited curves (including NIST's as well as femr2's) provide any real evidence against Chandler's claim. For the purpose of arguing against a near-instantaneous transition from rest to freefall, those curves are worse than useless: Their mathematical names and spurious detail can mislead even well-intentioned individuals into thinking those acceleration curves show something meaningful about the abruptness of the transition to freefall. They do not.

To demonstrate the uselessness of those particular acceleration curves for this particular purpose, I will introduce a counterexample in the form of yet another acceleration curve, motivated by this observation from part 1:

(In this graph, I've stuck with best-fit polynomials of degree n=2. As can be seen in the other graphs, raising the degree to n=3 has very little effect on the graph. That tells us the additional features seen when the degree is raised still further, to n=4, are probably artifacts of the smoothing process interacting with noise in the signal. It might even tell us more, but I'll save that for later.) ​
It's later now.

The fact that Savitzky-Golay differentiation and smoothing of femr2's position data for the northwest corner of WTC7 is essentially unaffected by raising the polynomial degree from 2 to 3 suggests the position data might be explained by piecewise constant (step) accelerations.

Consider, for example, the thick blue acceleration curve in this graph:

stepwise.png


That stepwise curve, if accurate, might support David Chandler's belief in a sudden transition from rest to free fall.

The red line is femr2's original "Savitzky-Golay" curve, which appears to have been derived by iterating the SG algorithm instead of computing the second derivative in one pass. (As explained in parts 1 and 2, those two processes are not equivalent.) That red curve, if accurate, might argue against David Chandler's claim.

Those are two very different curves, and you might think they have very different implications for Chandler's claim.

Those two curves are of comparable accuracy. The stepwise acceleration curve is a little more accurate overall, but its superior accuracy doesn't show up until 13 seconds and later (using femr2's time coordinates). Its accuracy between 12 and 13 seconds could be improved by moving the stepwise transitions slightly earlier in time, but I chose to use moderately round numbers:
Code:
         t < 12.4s            0 ft/s^2 (at rest)
12.4s <= t < 12.7s          -32 ft/s^2 (free fall)
12.7s <= t < 13.2s          -40 ft.s^2 (over-g)
13.2s <= t < 14.7s          -32 ft/s^2 (free fall)
14.7s <= t < 15.4s          -18 ft/s^2
15.4s <= t < 15.8s          -25 ft/s^2
15.8s <= t < 16.0s          -10 ft/s^2
16.0s <= t < 16.5s           -8 ft/s^2
16.5s <= t                  +12 ft/s^2

The thin green line shows the acceleration curve computed from femr2's position data using true Savitzky-Golay differentiation with best-fit polynomials of degree 2 and 61 data points per window.

The medium purple line looks like a smoothed version of the green line, but it was computed via true Savitzky-Golay differentiation from the positional data implied by the blue stepwise acceleration curve. (With piecewise constant accelerations, those positions can be computed analytically via double integration.)

That purple line approximates the green (SG) line remarkably well. Yet the purple line was derived from position data that were known to match a perfectly stepwise acceleration. That means the green (SG) line is quite similar to what we'd expect to get out of the Savitzky-Golay algorithm if femr2's position data were produced by perfectly instantaneous transitions between rest and free fall, et cetera. That means the green line, showing true Savitzky-Golay differentiation of the positional data femr2 extracted from the Dan Rather video, is consistent with David Chandler's claim.

If femr2 had provided us with an adequate description of the process he used to derive his original "Savitzky-Golay" curve (the red line), we could have applied his process to the positional data implied by the blue stepwise curve, thereby obtaining a similarly direct confirmation that femr2's original "Savitzky-Golay" curve is unable to rule out Chandler's claim. Since femr2 has expressed his delight in obstructing my attempts to understand his process, I will just say this:
  • The iterated SG femr2 appears to have used doubles the effective width of his convolution window while giving more weight to points in the center of that extended window than to points near its edges. (This is why femr2's process yields results that differ from true Savitzky-Golay differentiation and smoothing.)
  • The doubling of the effective window size worsens the lag and smearing problems I noted in part 2, while giving more weight to the window's central points tends to lessen those same problems.
  • Comparing femr2's original "Savitzky-Golay" curve (the red line) to a true Savitzky-Golay curve (the green line), it looks as though femr2's curve has smeared the onset of free fall over a longer period of time than has the true Savitzky-Golay curve with 61 points per window.
  • That conclusion is supported by the greater accuracy of the true Savitzky-Golay curve (as reported below).

Throughout this thread, many have lamented the absence of error analysis. We did some worst-case error analysis near the beginning of the thread, but those analyses addressed only the worst-case errors for a single point within a time series. So far as I recall, no one has ever posted an error analysis that compares the overall accuracy of different acceleration curves.

The most straightforward way to check the accuracy of an acceleration curve obtained by numerical differentiation from positional data is to use numerical integration to compute the positional data implied by the acceleration curve, and compare those computed positions to the actual positional data the acceleration curve purports to model.

For the three acceleration curves I've been discussing here, that error analysis yields these results:

error.png


Overall, femr2's original "Savitzky-Golay" curve is the least accurate of the three curves. The stepwise acceleration curve is a little more accurate. The true Savitzky-Golay acceleration curve is easily the most accurate of the three.

Does that mean the true Savitzky-Golay curve provides evidence against Chandler's claim that the transition from rest to free fall was abrupt? No. As shown by the purple curve in the first graph, Savitzky-Golay differentiation and smoothing will smear any abrupt changes in acceleration over a period of time that's roughly proportional to the window size. That means Savitzky-Golay acceleration curves are completely useless if your purpose is to refute Chandler's claim or to argue that NIST's estimated 2.25s of approximate free fall is incorrect.

Does that mean Chandler or NIST were right?

Not at all.

What it does mean is this: Those who have been using Savitzky-Golay curves to argue against Chandler or NIST have been relying upon a fallacious argument.
 
Last edited:
What it does mean is this: Those who have been using Savitzky-Golay curves to argue against Chandler or NIST have been relying upon a fallacious argument.

Thanks, W.D. I agree that S-G is a problematic approach. I appreciate pgimeno's cleaner work.

The error analysis is of course key to the understanding. Essentially, I argue that due to the errors inherent, Chandler has no basis to make a constant free fall claim. I would argue that there is also no basis to call the transition to free fall as instantaneous.
 
Very interesting an instructive, W.D., as usual.

One point, though:

I believe that the main source of error in femr2's graph is a lateral displacement of about 0.1 seconds. Look at this graph:

femr2-qtiplot-comparison.png


The displacement should be apparent. The n.5 (labeled) ticks in femr2's graph overlap with the n.4 (unlabeled) ticks in my graph.

Instead of doing two derivatives, as I did in my graph, and then two integrals after the smoothing, I directly compared the smoothed values with the raw values, obtaining this graph. That may induce an arbitrary constant offset with respect to your method (the integration constant), but that's no big deal. Here's the plot, with a zero line added:

SG1SG2minusRaw.png


(As a reminder, SG1 in these graphs means Savitzky-Golay smoothing with a symmetric window of 51 samples and polynomial degree 1, and SG2 ditto with degree 2. SG1&SG2 means applying one after the other. The order doesn't matter. The acceleration graph above comes from applying central differences twice on the result.)

The maximum and minimum are about 0.55 and -1.80 ft. respectively.

When displacing it by 0.1 seconds as in femr2's graph, I get a plot that starts increasing up to almost 13 ft, with certain similarity to your error plots.

That makes me think that either the raw data femr2 used does not match the one that he provides and from which we derive our graphs (because it uses a different t=0 reference), or he mislabeled the X axis. Or, more unlikely, he did something else with his data that resulted in a shift along the X axis.

In addition to the effects of SG smoothing, I wanted to mention that I have now looked at the Cam #3 data again, in relation to trying to determine whether the actual acceleration grew quickly or not. Cam #3 data is FAR less noisy, but it has issues with the unit conversion.

The following graph includes the velocity, as the derivative of the raw position data, and the acceleration. The vertical units are unspecified because I'm not sure what they are in femr2's data. Probably pixels. The horizontal units are frame number, which can be converted to time in seconds by dividing by 29.97. The "raw" velocity and acceleration use the axis on the left for units (albeit the axis is labeled units/frame, that's only for velocity; acceleration is in units/frame²). The "zoomed" acceleration's Y axis is the one to the right, also in units/frame². It has only one SG pass of degree 2 and with a 51 sample symmetric window.

Cam3_profiles.png


But it suffices to look at the velocity plot to see visually how the average slope changes gradually, not abruptly, in the interval between frame ~270 and frame ~300 (modulo the bumps). That velocity plot is not smoothed, so that's a powerful argument in favor of a gradual onset of acceleration.
 
I believe that the main source of error in femr2's graph is a lateral displacement of about 0.1 seconds. Look at this graph:

[qimg]http://www.formauri.es/personal/pgimeno/xfiles/11-s/femr2-qtiplot-comparison.png[/qimg]

The displacement should be apparent. The n.5 (labeled) ticks in femr2's graph overlap with the n.4 (unlabeled) ticks in my graph.

...snip...

When displacing [pgimeno's curve] by 0.1 seconds as in femr2's graph, I get a plot that starts increasing up to almost 13 ft, with certain similarity to your error plots.

That makes me think that either the raw data femr2 used does not match the one that he provides and from which we derive our graphs (because it uses a different t=0 reference), or he mislabeled the X axis. Or, more unlikely, he did something else with his data that resulted in a shift along the X axis.
I'm inclined to think you're using a data set that's shifted relative to the data set femr2 and I used. Here's why I think so: My own curves were derived directly from femr2's data (listed below), independently of femr2's red curve, yet the noise peaks you see at the right side of my error graph line up vertically with the noise peaks in femr2's red curve.

Here are the position data I've been using, extracted from femr2's DanRather1408.csv:


Code:
0.0 0.0
0.0167 -0.1523
0.0334 -0.0679
0.0501 -0.1759
0.0667 -0.1796
0.0834 -0.1735
0.1001 -0.235
0.1168 0.1338
0.1335 0.1138
0.1502 0.1166
0.1668 0.2275
0.1835 0.2453
0.2002 0.2898
0.2169 0.4384
0.2336 0.4211
0.2503 0.3658
0.2669 0.3391
0.2836 0.2813
0.3003 0.2739
0.317 0.2631
0.3337 0.2437
0.3504 0.2333
0.367 0.2261
0.3837 0.2279
0.4004 0.2162
0.4171 -0.0783
0.4338 -0.0422
0.4505 0.2515
0.4671 0.2564
0.4838 0.2713
0.5005 0.2663
0.5172 0.2662
0.5339 0.2339
0.5506 0.2146
0.5672 0.2504
0.5839 0.2959
0.6006 0.28
0.6173 0.2002
0.634 0.2541
0.6507 0.3201
0.6673 0.3276
0.684 0.3625
0.7007 0.4107
0.7174 0.4317
0.7341 0.3813
0.7508 0.3444
0.7674 0.3448
0.7841 0.4167
0.8008 0.4005
0.8175 -0.1187
0.8342 -0.0131
0.8509 0.1789
0.8675 0.095
0.8842 0.3318
0.9009 0.3044
0.9176 0.0739
0.9343 0.1038
0.951 -0.2258
0.9676 -0.17
0.9843 0.0175
1.001 -0.0627
1.0177 -0.2698
1.0344 -0.2821
1.0511 -0.2175
1.0677 -0.2244
1.0844 -0.2372
1.1011 -0.2689
1.1178 -0.0803
1.1345 -0.0545
1.1512 -0.3674
1.1678 -0.4078
1.1845 -0.3981
1.2012 -0.3391
1.2179 -0.3202
1.2346 -0.2445
1.2513 -0.158
1.2679 -0.1814
1.2846 -0.0753
1.3013 -0.0699
1.318 -0.3263
1.3347 -0.3036
1.3514 -0.3814
1.368 -0.3527
1.3847 -0.3067
1.4014 -0.3102
1.4181 -0.1936
1.4348 -0.1143
1.4515 -0.0554
1.4681 -0.0677
1.4848 -0.1694
1.5015 -0.1922
1.5182 -0.137
1.5349 -0.1069
1.5516 -0.1116
1.5682 -0.211
1.5849 -0.2092
1.6016 -0.2207
1.6183 -0.2316
1.635 -0.176
1.6517 -0.2192
1.6683 -0.2193
1.685 -0.119
1.7017 -0.0744
1.7184 -0.0438
1.7351 -0.0702
1.7518 0.004
1.7684 -0.0078
1.7851 -0.0621
1.8018 -0.0637
1.8185 0.0954
1.8352 0.0978
1.8519 -0.0591
1.8685 -0.1041
1.8852 0.0337
1.9019 -0.0171
1.9186 -0.3254
1.9353 -0.3636
1.952 -0.4961
1.9686 -0.4981
1.9853 -0.3012
2.002 -0.2925
2.0187 -0.1986
2.0354 -0.2387
2.0521 -0.1572
2.0687 -0.1
2.0854 -0.3211
2.1021 -0.3065
2.1188 -0.2307
2.1355 -0.2162
2.1522 -0.214
2.1688 -0.2106
2.1855 0.1188
2.2022 0.1148
2.2189 0.3266
2.2356 0.334
2.2523 0.6935
2.2689 0.6823
2.2856 0.5033
2.3023 0.4658
2.319 0.1368
2.3357 0.0812
2.3524 -0.3858
2.369 -0.423
2.3857 -0.0705
2.4024 -0.1565
2.4191 -0.318
2.4358 -0.353
2.4525 -0.0227
2.4691 0.065
2.4858 0.0524
2.5025 0.1294
2.5192 -0.2648
2.5359 -0.2392
2.5526 -0.0111
2.5692 -0.0241
2.5859 0.0368
2.6026 -0.0257
2.6193 0.0818
2.636 0.0883
2.6527 0.1474
2.6693 0.1958
2.686 0.2391
2.7027 0.2666
2.7194 -0.081
2.7361 -0.0264
2.7528 -0.5735
2.7694 -0.5558
2.7861 -0.7696
2.8028 -0.767
2.8195 -0.6983
2.8362 -0.7188
2.8529 -0.5475
2.8695 -0.5215
2.8862 -0.4922
2.9029 -0.4256
2.9196 -0.479
2.9363 -0.3006
2.953 -0.2741
2.9696 -0.8186
2.9863 -0.8528
3.003 -0.9743
3.0197 -1.0305
3.0364 -0.5558
3.0531 -0.5633
3.0697 -0.496
3.0864 -0.5612
3.1031 -0.508
3.1198 -0.3533
3.1365 -0.2574
3.1532 -0.2245
3.1698 -0.8324
3.1865 -0.7516
3.2032 -0.632
3.2199 -0.7079
3.2366 -0.6695
3.2533 -0.7404
3.2699 -0.4142
3.2866 -0.4406
3.3033 -0.4821
3.32 -0.4608
3.3367 -0.3567
3.3534 -0.2464
3.37 -0.3986
3.3867 -0.4372
3.4034 -0.4352
3.4201 -0.3746
3.4368 -0.2595
3.4535 -0.3417
3.4701 -0.3841
3.4868 -0.4322
3.5035 -0.4432
3.5202 -0.4705
3.5369 -0.4644
3.5536 -0.4799
3.5702 -0.3882
3.5869 -0.3357
3.6036 -0.5304
3.6203 -0.4711
3.637 -0.4444
3.6537 -0.4777
3.6703 -0.391
3.687 -0.3341
3.7037 -0.4342
3.7204 -0.5456
3.7371 -0.4929
3.7538 -0.4393
3.7704 -0.4146
3.7871 -0.4567
3.8038 -0.4249
3.8205 -0.3926
3.8372 -0.6015
3.8539 -0.5613
3.8705 -0.3452
3.8872 -0.4347
3.9039 -0.4356
3.9206 -0.4327
3.9373 -0.417
3.954 -0.2429
3.9706 -0.2077
3.9873 -0.3027
4.004 -0.2826
4.0207 -0.3262
4.0374 -0.8243
4.0541 -0.7914
4.0707 -0.2918
4.0874 -0.2904
4.1041 -0.3668
4.1208 -0.351
4.1375 -0.4939
4.1542 -0.5654
4.1708 -0.5779
4.1875 -0.7148
4.2042 -0.5759
4.2209 -0.495
4.2376 -0.4327
4.2543 -0.4546
4.2709 -0.6226
4.2876 -0.5787
4.3043 -0.64
4.321 -0.4819
4.3377 -0.5625
4.3544 -0.6621
4.371 -0.7064
4.3877 -0.6357
4.4044 -0.5112
4.4211 -0.6251
4.4378 -0.5338
4.4545 -0.5341
4.4711 -0.5789
4.4878 -0.5418
4.5045 -0.6262
4.5212 -0.6271
4.5379 -0.7062
4.5546 -0.4668
4.5712 -0.5057
4.5879 -0.6171
4.6046 -0.5664
4.6213 -0.6116
4.638 -0.5833
4.6547 -0.8846
4.6713 -0.8507
4.688 -0.6679
4.7047 -0.7657
4.7214 -0.6729
4.7381 -0.8211
4.7548 -0.244
4.7714 -0.2004
4.7881 -0.1129
4.8048 -0.1207
4.8215 -0.2661
4.8382 -0.2035
4.8549 -0.6398
4.8715 -0.6715
4.8882 -0.4097
4.9049 -0.371
4.9216 -0.1945
4.9383 -0.2161
4.955 -0.0556
4.9716 -0.0922
4.9883 -0.1505
5.005 -0.1908
5.0217 -0.1858
5.0384 -0.092
5.0551 -0.4472
5.0717 -0.5564
5.0884 -0.9412
5.1051 -0.7954
5.1218 -0.8022
5.1385 -0.7383
5.1552 -0.4025
5.1718 -0.4947
5.1885 -0.0646
5.2052 -0.1462
5.2219 -0.1571
5.2386 -0.1138
5.2553 -0.4186
5.2719 -0.4039
5.2886 -0.8827
5.3053 -0.7446
5.322 -0.6103
5.3387 -0.5149
5.3554 -0.5228
5.372 -0.5558
5.3887 -0.5682
5.4054 -0.6016
5.4221 -0.5723
5.4388 -0.5958
5.4555 -0.7204
5.4721 -0.6622
5.4888 -0.7108
5.5055 -0.7485
5.5222 -0.6613
5.5389 -0.5555
5.5556 -0.4519
5.5722 -0.3529
5.5889 -0.3862
5.6056 -0.4402
5.6223 -0.4546
5.639 -0.5131
5.6557 -0.5327
5.6723 -0.4812
5.689 -0.4527
5.7057 -0.3818
5.7224 -0.2988
5.7391 -0.3588
5.7558 -0.3505
5.7724 -0.491
5.7891 -0.4843
5.8058 -0.3666
5.8225 -0.4278
5.8392 -0.4525
5.8559 -0.4804
5.8725 -0.5119
5.8892 -0.4746
5.9059 -0.3912
5.9226 -0.4421
5.9393 -0.7823
5.956 -0.7243
5.9726 -0.3933
5.9893 -0.3832
6.006 -0.3673
6.0227 -0.4681
6.0394 -0.5872
6.0561 -0.6709
6.0727 -0.6535
6.0894 -0.6755
6.1061 -0.6858
6.1228 -0.6416
6.1395 -0.5968
6.1562 -0.5307
6.1728 -0.7395
6.1895 -0.7205
6.2062 -0.4904
6.2229 -0.4788
6.2396 -0.5455
6.2563 -0.561
6.2729 -0.5384
6.2896 -0.5467
6.3063 -0.5493
6.323 -0.5593
6.3397 -0.9136
6.3564 -0.8562
6.373 -0.9566
6.3897 -0.8626
6.4064 -0.8874
6.4231 -0.8167
6.4398 -0.9391
6.4565 -1.0201
6.4731 -0.9838
6.4898 -1.0157
6.5065 -1.0715
6.5232 -1.1053
6.5399 -0.8225
6.5566 -0.7682
6.5732 -0.9097
6.5899 -0.9125
6.6066 -0.8869
6.6233 -0.8388
6.64 -0.8522
6.6567 -0.8147
6.6733 -0.6685
6.69 -0.7325
6.7067 -0.842
6.7234 -0.9208
6.7401 -0.9466
6.7568 -0.9686
6.7734 -0.9106
6.7901 -0.8924
6.8068 -1.0611
6.8235 -1.0424
6.8402 -1.1127
6.8569 -1.0758
6.8735 -0.8865
6.8902 -0.8356
6.9069 -0.5785
6.9236 -0.5729
6.9403 -0.5829
6.957 -0.6301
6.9736 -0.522
6.9903 -0.5793
7.007 -0.9102
7.0237 -0.8766
7.0404 -0.8181
7.0571 -0.8354
7.0737 -0.8416
7.0904 -0.8117
7.1071 -0.7159
7.1238 -0.6208
7.1405 -0.607
7.1572 -0.5992
7.1738 -0.5543
7.1905 -0.5121
7.2072 -0.5677
7.2239 -0.5618
7.2406 -0.7467
7.2573 -0.7177
7.2739 -0.717
7.2906 -0.7453
7.3073 -0.6667
7.324 -0.674
7.3407 -0.5807
7.3574 -0.6642
7.374 -0.5934
7.3907 -0.644
7.4074 -0.8183
7.4241 -0.7727
7.4408 -0.7285
7.4575 -0.7593
7.4741 -0.741
7.4908 -0.7548
7.5075 -0.7943
7.5242 -0.8362
7.5409 -0.5512
7.5576 -0.5854
7.5742 -0.4846
7.5909 -0.4559
7.6076 -0.3867
7.6243 -0.3653
7.641 -0.3274
7.6577 -0.2086
7.6743 -0.0216
7.691 -0.0398
7.7077 -0.0706
7.7244 -0.0852
7.7411 -0.2325
7.7578 -0.3926
7.7744 -0.4253
7.7911 -0.3509
7.8078 -0.3627
7.8245 -0.3256
7.8412 -0.3302
7.8579 -0.3758
7.8745 -0.3736
7.8912 -0.4567
7.9079 -0.4419
7.9246 -0.3565
7.9413 -0.3383
7.958 -0.3711
7.9746 -0.3908
7.9913 -0.3737
8.008 -0.3395
8.0247 -0.402
8.0414 -0.6539
8.0581 -0.6286
8.0747 -0.5105
8.0914 -0.5115
8.1081 -0.407
8.1248 -0.323
8.1415 -0.2438
8.1582 -0.2831
8.1748 -0.4017
8.1915 -0.3063
8.2082 -0.191
8.2249 -0.2151
8.2416 -0.2761
8.2583 -0.3151
8.2749 -0.3558
8.2916 -0.3977
8.3083 -0.424
8.325 -0.4333
8.3417 -0.4051
8.3584 -0.3836
8.375 -0.4097
8.3917 -0.4228
8.4084 -0.4394
8.4251 -0.438
8.4418 -0.431
8.4585 -0.4156
8.4751 -0.3891
8.4918 -0.4297
8.5085 -0.4492
8.5252 -0.3883
8.5419 -0.332
8.5586 -0.2322
8.5752 -0.2751
8.5919 -0.1979
8.6086 -0.2912
8.6253 -0.1487
8.642 -0.1435
8.6587 -0.4271
8.6753 -0.3819
8.692 -0.4575
8.7087 -0.4443
8.7254 -0.3674
8.7421 -0.4384
8.7588 -0.5043
8.7754 -0.5546
8.7921 -0.5924
8.8088 -0.5901
8.8255 -0.49
8.8422 -0.5023
8.8589 -0.5565
8.8755 -0.5445
8.8922 -0.5186
8.9089 -0.4782
8.9256 -0.575
8.9423 -0.5668
8.959 -0.5321
8.9756 -0.5088
8.9923 -0.4959
9.009 -0.4879
9.0257 -0.544
9.0424 -0.5294
9.0591 -0.4294
9.0757 -0.442
9.0924 -0.453
9.1091 -0.456
9.1258 -0.4737
9.1425 -0.5101
9.1592 -0.5963
9.1758 -0.6431
9.1925 -0.4972
9.2092 -0.4849
9.2259 -0.6531
9.2426 -0.7061
9.2593 -0.7016
9.2759 -0.6671
9.2926 -0.673
9.3093 -0.6888
9.326 -0.6064
9.3427 -0.7141
9.3594 -0.5353
9.376 -0.5074
9.3927 -0.7278
9.4094 -0.717
9.4261 -0.7842
9.4428 -0.7609
9.4595 -0.7745
9.4761 -0.7724
9.4928 -0.5332
9.5095 -0.5911
9.5262 -0.4842
9.5429 -0.55
9.5596 -0.2828
9.5762 -0.259
9.5929 -0.5848
9.6096 -0.5784
9.6263 -0.4694
9.643 -0.4254
9.6597 -0.6587
9.6763 -0.6326
9.693 -0.6482
9.7097 -0.6957
9.7264 -0.6392
9.7431 -0.705
9.7598 -0.6513
9.7764 -0.5806
9.7931 -0.1598
9.8098 -0.2055
9.8265 -0.7234
9.8432 -0.8328
9.8599 -0.9307
9.8765 -0.8276
9.8932 -0.7884
9.9099 -0.7143
9.9266 -0.7223
9.9433 -0.7583
9.96 -0.5655
9.9766 -0.6015
9.9933 -0.5317
10.01 -0.5383
10.0267 -0.3572
10.0434 -0.3306
10.0601 -0.2262
10.0767 -0.2089
10.0934 -0.261
10.1101 -0.2789
10.1268 -0.2997
10.1435 -0.2334
10.1602 -0.3587
10.1768 -0.2763
10.1935 -0.2337
10.2102 -0.3729
10.2269 -0.3886
10.2436 -0.4756
10.2603 -0.4546
10.2769 -0.4584
10.2936 -0.4309
10.3103 -0.3819
10.327 -0.3738
10.3437 -0.3723
10.3604 -0.4111
10.377 -0.2413
10.3937 -0.2602
10.4104 -0.3917
10.4271 -0.3203
10.4438 -0.3026
10.4605 -0.1953
10.4771 -0.1582
10.4938 -0.1214
10.5105 -0.1009
10.5272 -0.1338
10.5439 -0.1623
10.5606 -0.1705
10.5772 -0.2289
10.5939 -0.3161
10.6106 -0.2043
10.6273 -0.2178
10.644 -0.284
10.6607 -0.2598
10.6773 -0.2459
10.694 -0.2547
10.7107 -0.4088
10.7274 -0.4248
10.7441 -0.408
10.7608 -0.3927
10.7774 -0.3683
10.7941 -0.3601
10.8108 -0.3193
10.8275 -0.2724
10.8442 -0.2646
10.8609 -0.2918
10.8775 -0.2662
10.8942 -0.2605
10.9109 -0.2903
10.9276 -0.252
10.9443 -0.2506
10.961 -0.2608
10.9776 -0.2299
10.9943 -0.2823
11.011 -0.2364
11.0277 -0.1455
11.0444 -0.1461
11.0611 -0.1133
11.0777 -0.0768
11.0944 -0.0816
11.1111 -0.127
11.1278 -0.1346
11.1445 -0.1736
11.1612 -0.1975
11.1778 -0.211
11.1945 -0.2186
11.2112 -0.2137
11.2279 -0.2686
11.2446 -0.2532
11.2613 -0.2011
11.2779 -0.1355
11.2946 -0.0454
11.3113 -0.1101
11.328 -0.8188
11.3447 -0.894
11.3614 -0.9957
11.378 -0.991
11.3947 -1.0228
11.4114 -1.0187
11.4281 -0.6166
11.4448 -0.6271
11.4615 -1.3696
11.4781 -1.3906
11.4948 -1.3655
11.5115 -1.3171
11.5282 -1.2252
11.5449 -1.309
11.5616 -1.1828
11.5782 -1.1636
11.5949 -1.2189
11.6116 -1.2042
11.6283 -0.9756
11.645 -0.8528
11.6617 -0.7886
11.6783 -0.8071
11.695 -0.8039
11.7117 -0.7454
11.7284 -0.9062
11.7451 -0.9438
11.7618 -0.8843
11.7784 -0.8169
11.7951 -0.6662
11.8118 -0.6756
11.8285 -0.3367
11.8452 -0.3864
11.8619 -0.4141
11.8785 -0.4646
11.8952 -0.7962
11.9119 -0.6555
11.9286 -0.8017
11.9453 -0.821
11.962 -0.754
11.9786 -0.8093
11.9953 -1.0116
12.012 -1.1058
12.0287 -0.9775
12.0454 -0.9356
12.0621 -0.922
12.0787 -0.9699
12.0954 -0.7152
12.1121 -0.6853
12.1288 -1.0212
12.1455 -1.017
12.1622 -0.9675
12.1788 -0.9282
12.1955 -1.0482
12.2122 -1.1149
12.2289 -1.1095
12.2456 -1.1243
12.2623 -0.7339
12.2789 -0.7697
12.2956 -1.2264
12.3123 -1.1409
12.329 -1.0986
12.3457 -1.173
12.3624 -1.1752
12.379 -1.2144
12.3957 -1.1458
12.4124 -1.1474
12.4291 -1.3284
12.4458 -1.3343
12.4625 -1.3838
12.4791 -1.3979
12.4958 -1.444
12.5125 -1.4065
12.5292 -1.3952
12.5459 -1.4145
12.5626 -1.5666
12.5792 -1.8953
12.5959 -1.9462
12.6126 -2.261
12.6293 -2.2875
12.646 -2.0186
12.6627 -2.059
12.6793 -2.9047
12.696 -3.1186
12.7127 -3.3596
12.7294 -3.4035
12.7461 -3.3943
12.7628 -3.7984
12.7794 -3.9399
12.7961 -4.5877
12.8128 -4.6994
12.8295 -4.763
12.8462 -5.1094
12.8629 -5.3097
12.8795 -5.6858
12.8962 -5.671
12.9129 -6.01
12.9296 -6.1547
12.9463 -6.4172
12.963 -7.3148
12.9796 -7.4753
12.9963 -8.0743
13.013 -9.2009
13.0297 -9.1942
13.0464 -9.3622
13.0631 -9.5265
13.0797 -9.8135
13.0964 -10.83
13.1131 -10.9248
13.1298 -11.3759
13.1465 -12.5176
13.1632 -12.6139
13.1798 -12.8545
13.1965 -13.9329
13.2132 -14.1435
13.2299 -14.536
13.2466 -15.52
13.2633 -15.774
13.2799 -16.3526
13.2966 -16.9459
13.3133 -17.2117
13.33 -18.1355
13.3467 -18.7648
13.3634 -19.0697
13.38 -20.2581
13.3967 -21.4364
13.4134 -21.7319
13.4301 -22.2129
13.4468 -23.0489
13.4635 -23.1168
13.4801 -23.7293
13.4968 -24.9475
13.5135 -25.2154
13.5302 -25.6247
13.5469 -26.738
13.5636 -27.7423
13.5802 -28.4334
13.5969 -29.1841
13.6136 -29.9538
13.6303 -30.202
13.647 -30.8259
13.6637 -32.0253
13.6803 -32.6372
13.697 -33.1152
13.7137 -34.2807
13.7304 -35.5029
13.7471 -35.9056
13.7638 -36.3789
13.7804 -37.7017
13.7971 -39.0319
13.8138 -39.5042
13.8305 -39.892
13.8472 -41.0238
13.8639 -42.2547
13.8805 -43.1674
13.8972 -43.5803
13.9139 -44.5486
13.9306 -46.0312
13.9473 -46.9655
13.964 -47.313
13.9806 -48.1299
13.9973 -49.4958
14.014 -50.3215
14.0307 -51.2172
14.0474 -52.2878
14.0641 -53.3967
14.0807 -54.6025
14.0974 -55.9323
14.1141 -57.0028
14.1308 -57.3971
14.1475 -58.2515
14.1642 -59.5814
14.1808 -60.8045
14.1975 -62.2402
14.2142 -62.9847
14.2309 -63.7162
14.2476 -65.0913
14.2643 -66.3647
14.2809 -67.3813
14.2976 -67.999
14.3143 -69.1592
14.331 -70.6337
14.3477 -72.0725
14.3644 -73.5631
14.381 -74.9158
14.3977 -75.7054
14.4144 -76.2401
14.4311 -77.496
14.4478 -79.1317
14.4645 -80.5244
14.4811 -81.8142
14.4978 -82.9617
14.5145 -84.0464
14.5312 -85.0169
14.5479 -86.1444
14.5646 -87.6418
14.5812 -89.1559
14.5979 -90.6325
14.6146 -91.8901
14.6313 -92.723
14.648 -93.4641
14.6647 -94.7437
14.6813 -96.3208
14.698 -97.8934
14.7147 -99.3399
14.7314 -100.8951
14.7481 -102.5128
14.7648 -103.9798
14.7814 -105.4086
14.7981 -106.6913
14.8148 -107.684
14.8315 -108.4888
14.8482 -109.8071
14.8649 -111.5422
14.8815 -113.2321
14.8982 -114.8515
14.9149 -116.4303
14.9316 -117.9264
14.9483 -119.2255
14.965 -120.3291
14.9816 -121.591
14.9983 -122.7242
15.015 -123.8802
15.0317 -125.5571
15.0484 -127.3223
15.0651 -128.9826
15.0817 -130.561
15.0984 -132.044
15.1151 -133.4949
15.1318 -134.8585
15.1485 -136.2618
15.1652 -137.5653
15.1818 -138.4751
15.1985 -139.6255
15.2152 -141.2207
15.2319 -142.8411
15.2486 -144.5246
15.2653 -146.2004
15.2819 -147.6579
15.2986 -149.0572
15.3153 -150.5823
15.332 -152.1786
15.3487 -154.0237
15.3654 -155.7683
15.382 -157.4439
15.3987 -159.1024
15.4154 -160.6389
15.4321 -162.2908
15.4488 -163.9234
15.4655 -165.29
15.4821 -166.4323
15.4988 -167.9236
15.5155 -169.7961
15.5322 -171.191
15.5489 -172.5795
15.5656 -173.847
15.5822 -175.4405
15.5989 -177.12
15.6156 -178.6147
15.6323 -180.2599
15.649 -181.8616
15.6657 -183.6436
15.6823 -185.3123
15.699 -187.0351
15.7157 -188.7603
15.7324 -190.5132
15.7491 -192.2485
15.7658 -194.0437
15.7824 -195.7613
15.7991 -197.4714
15.8158 -199.2176
15.8325 -200.9825
15.8492 -202.7504
15.8659 -204.4663
15.8825 -206.1556
15.8992 -207.8476
15.9159 -209.5047
15.9326 -211.2423
15.9493 -213.0621
15.966 -214.7932
15.9826 -216.596
15.9993 -218.3177
16.016 -219.9446
16.0327 -221.7472
16.0494 -223.4839
16.0661 -225.1681
16.0827 -226.9753
16.0994 -228.7342
16.1161 -230.4206
16.1328 -232.0275
16.1495 -233.6789
16.1662 -235.5008
16.1828 -237.3433
16.1995 -239.053
16.2162 -240.7799
16.2329 -242.5891
16.2496 -244.4542
16.2663 -246.0778
16.2829 -247.6154
16.2996 -249.3602
16.3163 -251.2078
16.333 -253.0346
16.3497 -254.7268
16.3664 -256.7536
16.383 -258.6549
16.3997 -260.451
16.4164 -262.4892
16.4331 -264.4854
16.4498 -266.3332
16.4665 -268.2011
16.4831 -269.9291
16.4998 -271.5394
16.5165 -273.3185
16.5332 -275.2149
16.5499 -276.9441
16.5666 -278.7201
16.5832 -280.427
16.5999 -282.1348
16.6166 -283.9295
16.6333 -285.5659
16.65 -287.4155
16.6667 -289.1887
16.6834 -290.9082
16.7 -292.7084
16.7167 -294.3614
16.7334 -296.1074
16.7501 -298.0014
16.7668 -299.7605
16.7835 -301.4422
16.8001 -303.2189
16.8168 -305.0457
16.8335 -306.7972
16.8502 -308.4766
16.8669 -310.1751
16.8836 -311.9285
16.9002 -313.7128
16.9169 -315.5522
16.9336 -317.251
16.9503 -319.0126
16.967 -320.8865
16.9837 -322.6138
17.0003 -324.0783
17.017 -325.6097
17.0337 -327.3137
17.0504 -329.012
17.0671 -330.3051
17.0838 -331.56
17.1004 -333.2652
17.1171 -334.7954
17.1338 -336.3737
17.1505 -337.9062
17.1672 -339.2303
17.1839 -340.6705


For reference, here is my digitization of femr2's original "Savitzky-Golay" acceleration curve (the red curve):


Code:
11.0  -4.386792452830189
11.006864988558352  -4.445754716981133
11.013729977116705  -4.504716981132077
11.020594965675057  -4.6226415094339615
11.027459954233409  -4.681603773584905
11.034324942791763  -4.681603773584905
11.041189931350115  -4.681603773584905
11.048054919908466  -4.681603773584905
11.05491990846682  -4.681603773584905
11.061784897025172  -4.681603773584905
11.068649885583524  -4.5636792452830175
11.075514874141877  -4.327830188679243
11.082379862700229  -4.150943396226415
11.08924485125858  -3.974056603773585
11.096109839816934  -3.738207547169811
11.102974828375286  -3.4433962264150946
11.109839816933638  -3.2075471698113205
11.116704805491992  -2.9716981132075464
11.123569794050344  -2.794811320754718
11.130434782608695  -2.5
11.137299771167047  -2.264150943396226
11.1441647597254  -1.9103773584905657
11.151029748283753  -1.6155660377358494
11.157894736842104  -1.084905660377359
11.164759725400458  -0.9080188679245271
11.17162471395881  -0.7311320754716988
11.178489702517162  -0.4952830188679247
11.185354691075515  -0.2594339622641506
11.192219679633867  0.09433962264150964
11.199084668192219  0.3891509433962259
11.205949656750573  0.625
11.212814645308924  0.8608490566037741
11.219679633867276  1.0966981132075464
11.22654462242563  1.3325471698113205
11.233409610983982  1.5094339622641506
11.240274599542333  1.5683962264150946
11.247139588100687  1.6273584905660385
11.254004576659039  1.7452830188679247
11.26086956521739  1.8632075471698109
11.267734553775744  1.9221698113207548
11.274599542334096  1.9221698113207548
11.281464530892448  2.040094339622641
11.288329519450802  2.040094339622641
11.295194508009153  2.040094339622641
11.302059496567505  2.099056603773585
11.308924485125859  2.216981132075472
11.31578947368421  2.334905660377359
11.322654462242562  2.393867924528301
11.329519450800916  2.5117924528301883
11.336384439359268  2.6886792452830193
11.34324942791762  2.7476415094339615
11.350114416475973  2.8655660377358485
11.356979405034325  2.8655660377358485
11.363844393592677  2.8066037735849054
11.37070938215103  2.7476415094339615
11.377574370709382  2.6886792452830193
11.384439359267734  2.5117924528301883
11.391304347826086  2.275943396226415
11.39816933638444  2.099056603773585
11.405034324942791  1.9221698113207548
11.411899313501145  1.6863207547169807
11.418764302059497  1.3915094339622645
11.425629290617849  1.0377358490566042
11.4324942791762  0.8608490566037741
11.439359267734554  0.625
11.446224256292906  0.33018867924528195
11.453089244851258  -0.023584905660378297
11.459954233409611  -0.2594339622641506
11.466819221967963  -0.4952830188679247
11.473684210526315  -0.6132075471698109
11.480549199084669  -0.7311320754716988
11.48741418764302  -0.8490566037735849
11.494279176201372  -0.9080188679245271
11.501144164759726  -0.9080188679245271
11.508009153318078  -0.9080188679245271
11.51487414187643  -0.9080188679245271
11.521739130434783  -0.9080188679245271
11.528604118993135  -0.9080188679245271
11.535469107551487  -0.9669811320754729
11.54233409610984  -1.025943396226415
11.549199084668192  -1.025943396226415
11.556064073226544  -1.025943396226415
11.562929061784898  -1.084905660377359
11.56979405034325  -1.2028301886792452
11.576659038901601  -1.3207547169811331
11.583524027459955  -1.4386792452830193
11.590389016018307  -1.4976415094339615
11.597254004576659  -1.5566037735849054
11.604118993135012  -1.6155660377358494
11.610983981693364  -1.6155660377358494
11.617848970251716  -1.6155660377358494
11.62471395881007  -1.6155660377358494
11.631578947368421  -1.5566037735849054
11.638443935926773  -1.3797169811320753
11.645308924485127  -1.1438679245283012
11.652173913043478  -0.9080188679245271
11.65903890160183  -0.6721698113207548
11.665903890160184  -0.4952830188679247
11.672768878718536  -0.3183962264150946
11.679633867276888  -0.08254716981132049
11.68649885583524  0.2122641509433958
11.693363844393593  0.4481132075471699
11.700228832951945  0.683962264150944
11.707093821510297  0.8018867924528301
11.71395881006865  0.9787735849056602
11.720823798627002  1.2146226415094343
11.727688787185354  1.4504716981132066
11.734553775743708  1.6273584905660385
11.74141876430206  1.6863207547169807
11.748283752860411  1.7452830188679247
11.755148741418765  1.9221698113207548
11.762013729977117  2.099056603773585
11.768878718535468  2.275943396226415
11.775743707093822  2.452830188679245
11.782608695652174  2.5707547169811322
11.789473684210526  2.6297169811320753
11.79633867276888  2.6886792452830193
11.803203661327231  2.7476415094339615
11.810068649885583  2.7476415094339615
11.816933638443937  2.8066037735849054
11.823798627002288  2.7476415094339615
11.83066361556064  2.7476415094339615
11.837528604118994  2.7476415094339615
11.844393592677346  2.7476415094339615
11.851258581235697  2.7476415094339615
11.858123569794051  2.7476415094339615
11.864988558352403  2.8066037735849054
11.871853546910755  2.9245283018867925
11.878718535469108  3.1014150943396217
11.88558352402746  3.337264150943396
11.892448512585812  3.573113207547169
11.899313501144164  3.75
11.906178489702517  3.808962264150943
11.91304347826087  3.808962264150943
11.919908466819223  3.92688679245283
11.926773455377575  3.92688679245283
11.933638443935926  3.867924528301887
11.940503432494278  3.808962264150943
11.947368421052632  3.691037735849056
11.954233409610984  3.3962264150943398
11.96796338672769  3.0424528301886795
11.974828375286041  2.8655660377358485
11.981693363844393  2.5707547169811322
11.988558352402746  2.216981132075472
11.995423340961098  1.8632075471698109
12.00228832951945  1.3915094339622645
12.009153318077804  0.919811320754718
12.016018306636155  0.683962264150944
12.022883295194507  0.3891509433962259
12.02974828375286  -0.023584905660378297
12.036613272311213  -0.4952830188679247
12.043478260869565  -0.9669811320754729
12.050343249427918  -1.2028301886792452
12.05720823798627  -1.4976415094339615
12.064073226544622  -1.9103773584905657
12.070938215102975  -2.382075471698114
12.077803203661327  -2.853773584905662
12.084668192219679  -3.089622641509436
12.091533180778033  -3.4433962264150946
12.098398169336384  -3.915094339622641
12.105263157894736  -4.386792452830189
12.11212814645309  -4.8584905660377355
12.118993135011442  -5.44811320754717
12.125858123569794  -5.919811320754718
12.132723112128147  -6.15566037735849
12.139588100686499  -6.450471698113208
12.14645308924485  -6.745283018867923
12.153318077803203  -7.216981132075471
12.160183066361556  -7.570754716981131
12.167048054919908  -7.7476415094339615
12.173913043478262  -7.924528301886792
12.180778032036613  -8.04245283018868
12.187643020594965  -8.160377358490564
12.194508009153317  -8.278301886792452
12.20137299771167  -8.39622641509434
12.208237986270023  -8.514150943396228
12.215102974828376  -8.808962264150946
12.235697940503432  -9.339622641509436
12.242562929061785  -9.81132075471698
12.249427917620137  -10.047169811320757
12.25629290617849  -10.341981132075471
12.263157894736842  -10.63679245283019
12.270022883295194  -10.99056603773585
12.276887871853546  -11.226415094339622
12.2837528604119  -11.462264150943398
12.290617848970252  -11.52122641509434
12.297482837528603  -11.580188679245282
12.304347826086957  -11.69811320754717
12.311212814645309  -11.757075471698112
12.31807780320366  -11.816037735849054
12.324942791762014  -11.875
12.331807780320366  -11.933962264150946
12.338672768878718  -12.05188679245283
12.345537757437071  -12.22877358490566
12.352402745995423  -12.523584905660378
12.359267734553775  -12.877358490566039
12.366132723112129  -13.054245283018869
12.37299771167048  -13.290094339622641
12.379862700228832  -13.525943396226413
12.386727688787186  -13.93867924528302
12.393592677345538  -14.29245283018868
12.40045766590389  -14.410377358490564
12.407322654462243  -14.64622641509434
12.414187643020595  -14.882075471698112
12.421052631578947  -15.0
12.4279176201373  -15.235849056603776
12.434782608695652  -15.35377358490566
12.441647597254004  -15.471698113207548
12.448512585812356  -15.707547169811324
12.45537757437071  -15.943396226415093
12.462242562929061  -16.179245283018872
12.469107551487415  -16.41509433962264
12.475972540045767  -16.88679245283019
12.482837528604119  -17.122641509433965
12.48970251716247  -17.358490566037734
12.496567505720824  -17.712264150943394
12.503432494279176  -18.30188679245283
12.51029748283753  -19.009433962264154
12.517162471395881  -19.245283018867923
12.524027459954233  -19.4811320754717
12.530892448512585  -19.952830188679247
12.537757437070939  -20.424528301886795
12.54462242562929  -20.89622641509434
12.551487414187644  -20.89622641509434
12.558352402745996  -21.191037735849058
12.565217391304348  -21.42688679245283
12.5720823798627  -21.60377358490566
12.578947368421053  -21.780660377358487
12.585812356979405  -22.016509433962263
12.592677345537757  -22.25235849056604
12.59954233409611  -22.429245283018872
12.606407322654462  -22.6061320754717
12.613272311212814  -22.841981132075468
12.620137299771168  -23.136792452830186
12.62700228832952  -23.608490566037737
12.633867276887871  -23.844339622641513
12.640732265446225  -24.139150943396224
12.647597254004577  -24.551886792452827
12.654462242562929  -25.02358490566038
12.661327231121282  -25.495283018867923
12.668192219679634  -25.7311320754717
12.675057208237986  -26.025943396226417
12.68192219679634  -26.43867924528302
12.688787185354691  -27.028301886792455
12.695652173913043  -27.61792452830189
12.702517162471395  -28.20754716981132
12.709382151029748  -28.797169811320757
12.7162471395881  -29.150943396226417
12.723112128146454  -29.504716981132077
12.729977116704806  -30.212264150943398
12.736842105263158  -30.801886792452827
12.74370709382151  -31.391509433962263
12.750572082379863  -31.62735849056604
12.757437070938215  -31.863207547169814
12.764302059496568  -32.33490566037736
12.77116704805492  -32.68867924528302
12.778032036613272  -33.160377358490564
12.784897025171624  -33.39622641509434
12.791762013729977  -33.69103773584906
12.79862700228833  -34.10377358490566
12.805491990846681  -34.575471698113205
12.812356979405035  -35.16509433962264
12.819221967963387  -35.56603773584905
12.826086956521738  -36.10849056603774
12.832951945080092  -36.22641509433962
12.839816933638444  -36.52122641509434
12.846681922196797  -36.75707547169811
12.85354691075515  -36.99292452830189
12.860411899313501  -37.22877358490566
12.867276887871853  -37.40566037735849
12.874141876430206  -37.52358490566038
12.881006864988558  -37.64150943396226
12.88787185354691  -37.818396226415096
12.894736842105264  -38.05424528301887
12.901601830663616  -38.2311320754717
12.908466819221967  -38.29009433962264
12.915331807780321  -38.34905660377358
12.922196796338673  -38.46698113207547
12.929061784897025  -38.58490566037736
12.935926773455378  -38.6438679245283
12.94279176201373  -38.6438679245283
12.949656750572082  -38.6438679245283
12.956521739130435  -38.58490566037736
12.963386727688787  -38.40801886792453
12.97025171624714  -38.17216981132076
12.977116704805493  -37.93632075471698
12.983981693363845  -37.759433962264154
12.990846681922196  -37.58254716981132
12.997711670480548  -37.52358490566038
13.004576659038902  -37.40566037735849
13.011441647597254  -37.346698113207545
13.018306636155607  -37.2877358490566
13.025171624713959  -37.346698113207545
13.03203661327231  -37.464622641509436
13.038901601830663  -37.700471698113205
13.045766590389016  -37.87735849056604
13.052631578947368  -37.93632075471698
13.059496567505722  -38.113207547169814
13.066361556064074  -38.34905660377358
13.073226544622425  -38.52594339622642
13.080091533180777  -38.70283018867924
13.08695652173913  -38.761792452830186
13.093821510297483  -38.761792452830186
13.100686498855836  -38.761792452830186
13.107551487414188  -38.761792452830186
13.11441647597254  -38.70283018867924
13.121281464530892  -38.58490566037736
13.128146453089245  -38.46698113207547
13.135011441647597  -38.29009433962264
13.14187643020595  -38.2311320754717
13.148741418764303  -38.05424528301887
13.155606407322654  -37.818396226415096
13.162471395881006  -37.64150943396226
13.16933638443936  -37.52358490566038
13.176201372997712  -37.464622641509436
13.183066361556063  -37.40566037735849
13.189931350114417  -37.2877358490566
13.196796338672769  -37.16981132075472
13.20366132723112  -37.110849056603776
13.210526315789473  -37.110849056603776
13.217391304347826  -37.05188679245283
13.22425629290618  -37.110849056603776
13.231121281464532  -37.16981132075472
13.237986270022883  -37.22877358490566
13.244851258581235  -37.22877358490566
13.251716247139587  -37.22877358490566
13.25858123569794  -37.22877358490566
13.265446224256292  -37.16981132075472
13.272311212814646  -36.99292452830189
13.279176201372998  -36.69811320754717
13.28604118993135  -36.52122641509434
13.292906178489702  -36.34433962264151
13.299771167048055  -36.22641509433962
13.306636155606407  -36.049528301886795
13.31350114416476  -35.87264150943396
13.320366132723112  -35.75471698113208
13.327231121281464  -35.636792452830186
13.334096109839816  -35.5188679245283
13.34096109839817  -35.45990566037736
13.347826086956522  -35.45990566037736
13.354691075514875  -35.40094339622642
13.361556064073227  -35.28301886792453
13.368421052631579  -35.16509433962264
13.37528604118993  -35.04716981132076
13.382151029748284  -34.87028301886792
13.389016018306636  -34.693396226415096
13.39588100686499  -34.575471698113205
13.402745995423341  -34.45754716981132
13.409610983981693  -34.39858490566038
13.416475972540045  -34.45754716981132
13.423340961098397  -34.51650943396226
13.43020594965675  -34.575471698113205
13.437070938215104  -34.75235849056604
13.443935926773456  -34.92924528301887
13.45766590389016  -35.34198113207547
13.464530892448513  -35.45990566037736
13.471395881006865  -35.75471698113208
13.478260869565217  -36.10849056603774
13.48512585812357  -36.285377358490564
13.491990846681922  -36.4622641509434
13.498855835240274  -36.58018867924528
13.505720823798626  -36.69811320754717
13.51258581235698  -36.75707547169811
13.519450800915331  -36.75707547169811
13.526315789473685  -36.75707547169811
13.533180778032037  -36.639150943396224
13.540045766590389  -36.34433962264151
13.54691075514874  -36.10849056603774
13.553775743707094  -35.636792452830186
13.560640732265446  -35.04716981132076
13.5675057208238  -34.45754716981132
13.574370709382151  -33.985849056603776
13.581235697940503  -33.59276729559748
13.588100686498855  -33.3372641509434
13.594965675057209  -33.21933962264151
13.60183066361556  -33.160377358490564
13.608695652173914  -33.21933962264151
13.615560640732266  -33.160377358490564
13.62929061784897  -33.514150943396224
13.636155606407323  -33.45518867924528
13.643020594965675  -33.45518867924528
13.649885583524028  -33.45518867924528
13.65675057208238  -33.45518867924528
13.670480549199084  -33.10141509433962
13.677345537757438  -32.57075471698113
13.68421052631579  -32.45283018867924
13.691075514874143  -32.45283018867924
13.697940503432495  -32.45283018867924
13.704805491990847  -32.3938679245283
13.711670480549198  -32.33490566037736
13.766590389016018  -31.922169811320757
13.77345537757437  -31.863207547169814
13.780320366132724  -31.745283018867923
13.787185354691076  -31.62735849056604
13.794050343249427  -31.568396226415096
13.80091533180778  -31.62735849056604
13.807780320366133  -31.68632075471698
13.814645308924485  -31.68632075471698
13.821510297482838  -32.3938679245283
13.82837528604119  -32.57075471698113
13.835240274599542  -32.806603773584904
13.842105263157894  -33.10141509433962
13.848970251716247  -33.41981132075471
13.8558352402746  -33.86792452830189
13.862700228832953  -34.10377358490566
13.869565217391305  -34.575471698113205
13.876430205949656  -34.634433962264154
13.883295194508008  -34.693396226415096
13.890160183066362  -34.81132075471698
13.897025171624714  -34.87028301886792
13.903890160183067  -34.87028301886792
13.91075514874142  -34.87028301886792
13.917620137299771  -34.81132075471698
13.924485125858123  -34.75235849056604
13.931350114416476  -34.51650943396226
13.938215102974828  -34.1627358490566
13.94508009153318  -33.985849056603776
13.951945080091534  -33.63207547169811
13.958810068649885  -33.3372641509434
13.965675057208237  -33.04245283018868
13.972540045766591  -32.806603773584904
13.979405034324943  -32.57075471698113
13.986270022883296  -32.45283018867924
13.993135011441648  -32.3938679245283
14.0  -32.3938679245283
14.006864988558352  -32.45283018867924
14.013729977116704  -32.62971698113208
14.020594965675057  -32.924528301886795
14.027459954233409  -33.278301886792455
14.034324942791763  -33.45518867924528
14.041189931350115  -33.63207547169811
14.048054919908466  -33.75
14.054919908466818  -33.80896226415094
14.061784897025172  -33.80896226415094
14.068649885583524  -33.75
14.075514874141877  -33.69103773584906
14.082379862700229  -33.57311320754717
14.08924485125858  -33.278301886792455
14.096109839816933  -33.04245283018868
14.102974828375286  -32.74764150943396
14.10983981693364  -32.57075471698113
14.116704805491992  -32.45283018867924
14.123569794050344  -32.3938679245283
14.130434782608695  -32.33490566037736
14.1441647597254  -32.33490566037736
14.151029748283753  -32.3938679245283
14.157894736842106  -32.3938679245283
14.164759725400458  -32.45283018867924
14.17162471395881  -32.62971698113208
14.178489702517162  -32.865566037735846
14.185354691075514  -33.04245283018868
14.192219679633867  -33.160377358490564
14.199084668192219  -33.278301886792455
14.205949656750573  -33.45518867924528
14.212814645308924  -33.69103773584906
14.219679633867276  -33.86792452830189
14.22654462242563  -33.92688679245283
14.233409610983982  -33.92688679245283
14.240274599542335  -33.92688679245283
14.247139588100687  -33.86792452830189
14.254004576659039  -33.69103773584906
14.26086956521739  -33.45518867924528
14.267734553775744  -33.278301886792455
14.274599542334096  -33.160377358490564
14.28146453089245  -32.924528301886795
14.288329519450802  -32.68867924528302
14.295194508009153  -32.45283018867924
14.302059496567505  -31.9811320754717
14.308924485125857  -31.922169811320757
14.31578947368421  -31.745283018867923
14.322654462242562  -31.509433962264154
14.329519450800916  -31.155660377358487
14.336384439359268  -30.860849056603776
14.34324942791762  -30.389150943396224
14.350114416475972  -30.035377358490564
14.356979405034325  -29.858490566037737
14.363844393592677  -29.62264150943396
14.37070938215103  -29.504716981132077
14.377574370709382  -29.445754716981128
14.384439359267734  -29.504716981132077
14.391304347826086  -29.681603773584904
14.39816933638444  -30.182783018867923
14.405034324942791  -30.683962264150942
14.411899313501145  -31.037735849056602
14.418764302059497  -31.214622641509436
14.425629290617849  -31.450471698113205
14.4324942791762  -31.68632075471698
14.439359267734554  -31.863207547169814
14.446224256292906  -31.922169811320757
14.45308924485126  -31.922169811320757
14.459954233409611  -31.922169811320757
14.466819221967963  -31.863207547169814
14.473684210526315  -31.804245283018872
14.480549199084667  -31.745283018867923
14.48741418764302  -31.62735849056604
14.494279176201372  -31.568396226415096
14.501144164759726  -31.509433962264154
14.508009153318078  -31.33254716981132
14.51487414187643  -31.096698113207545
14.521739130434783  -30.801886792452827
14.528604118993135  -30.566037735849058
14.535469107551489  -30.212264150943398
14.54233409610984  -29.91745283018868
14.549199084668192  -29.62264150943396
14.556064073226544  -29.386792452830186
14.562929061784898  -28.91509433962264
14.56979405034325  -28.443396226415096
14.576659038901603  -28.14858490566038
14.583524027459955  -27.971698113207545
14.590389016018307  -27.794811320754718
14.597254004576659  -27.028301886792455
14.60411899313501  -26.910377358490564
14.610983981693364  -26.85141509433962
14.617848970251716  -26.674528301886795
14.62471395881007  -26.43867924528302
14.631578947368421  -26.202830188679243
14.638443935926773  -25.849056603773583
14.645308924485125  -25.613207547169814
14.652173913043478  -25.259433962264154
14.65903890160183  -24.964622641509436
14.665903890160184  -24.669811320754718
14.672768878718536  -24.433962264150942
14.679633867276888  -24.05660377358491
14.68649885583524  -23.579009433962263
14.693363844393593  -23.31367924528302
14.700228832951945  -23.0188679245283
14.707093821510298  -22.783018867924532
14.71395881006865  -22.429245283018872
14.720823798627002  -22.25235849056604
14.727688787185354  -22.075471698113205
14.734553775743708  -21.89858490566038
14.74141876430206  -21.662735849056606
14.748283752860413  -21.485849056603776
14.755148741418765  -21.367924528301888
14.762013729977117  -21.308962264150942
14.768878718535468  -21.25
14.77574370709382  -21.308962264150942
14.782608695652174  -21.42688679245283
14.789473684210526  -21.662735849056606
14.79633867276888  -22.075471698113205
14.803203661327231  -22.31132075471698
14.810068649885583  -22.547169811320757
14.816933638443935  -23.183962264150942
14.823798627002288  -23.844339622641513
14.830663615560642  -24.198113207547173
14.837528604118994  -24.375
14.844393592677346  -24.610849056603776
14.851258581235697  -24.787735849056602
14.858123569794051  -24.905660377358487
14.864988558352403  -24.964622641509436
14.871853546910756  -24.846698113207545
14.878718535469108  -24.610849056603776
14.88558352402746  -24.433962264150942
14.892448512585812  -24.25707547169811
14.899313501144164  -23.962264150943398
14.906178489702517  -23.72641509433962
14.91304347826087  -23.254716981132077
14.919908466819223  -22.95990566037736
14.926773455377575  -22.783018867924532
14.933638443935926  -22.6061320754717
14.940503432494278  -22.370283018867923
14.947368421052632  -22.193396226415096
14.954233409610984  -22.075471698113205
14.961098398169337  -22.016509433962263
14.96796338672769  -22.016509433962263
14.974828375286041  -22.016509433962263
14.981693363844393  -22.016509433962263
14.988558352402746  -22.075471698113205
14.995423340961098  -22.134433962264154
15.002288329519452  -22.134433962264154
15.009153318077804  -22.134433962264154
15.016018306636155  -22.134433962264154
15.022883295194507  -21.957547169811324
15.02974828375286  -21.721698113207548
15.036613272311213  -21.320754716981135
15.043478260869566  -20.955188679245282
15.050343249427918  -20.660377358490564
15.05720823798627  -20.424528301886795
15.064073226544622  -19.83490566037736
15.070938215102974  -19.36320754716981
15.077803203661327  -18.950471698113205
15.084668192219679  -18.655660377358487
15.091533180778033  -18.419811320754718
15.098398169336384  -18.125
15.105263157894736  -17.94811320754717
15.112128146453088  -17.889150943396224
15.118993135011442  -17.830188679245282
15.125858123569794  -17.889150943396224
15.132723112128147  -17.94811320754717
15.139588100686499  -18.125
15.14645308924485  -18.30188679245283
15.153318077803204  -18.419811320754718
15.160183066361556  -18.537735849056606
15.16704805491991  -18.655660377358487
15.173913043478262  -18.77358490566038
15.180778032036613  -19.245283018867923
15.187643020594965  -19.245283018867923
15.194508009153317  -19.304245283018872
15.20137299771167  -19.304245283018872
15.208237986270023  -19.304245283018872
15.215102974828376  -19.304245283018872
15.221967963386728  -19.18632075471698
15.22883295194508  -18.891509433962266
15.235697940503432  -18.714622641509436
15.242562929061785  -18.47877358490566
15.249427917620137  -18.183962264150942
15.25629290617849  -17.94811320754717
15.263157894736842  -17.594339622641513
15.270022883295194  -17.122641509433965
15.276887871853546  -16.827830188679247
15.2837528604119  -16.59198113207547
15.290617848970252  -16.3561320754717
15.297482837528605  -16.179245283018872
15.304347826086957  -16.06132075471698
15.31807780320366  -15.707547169811324
15.324942791762014  -15.648584905660378
15.331807780320366  -15.648584905660378
15.33867276887872  -15.707547169811324
15.345537757437071  -16.10062893081761
15.352402745995423  -16.3561320754717
15.359267734553775  -16.474056603773583
15.366132723112127  -16.88679245283019
15.37299771167048  -17.47641509433962
15.379862700228832  -18.066037735849058
15.386727688787186  -18.30188679245283
15.393592677345538  -18.655660377358487
15.40045766590389  -19.150943396226413
15.407322654462241  -19.71698113207547
15.414187643020595  -20.070754716981128
15.421052631578947  -20.247641509433965
15.4279176201373  -20.483490566037734
15.434782608695652  -20.719339622641513
15.441647597254004  -21.014150943396224
15.448512585812358  -21.485849056603776
15.45537757437071  -21.957547169811324
15.462242562929063  -22.547169811320757
15.469107551487415  -22.783018867924532
15.475972540045767  -23.136792452830186
15.482837528604119  -23.72641509433962
15.48970251716247  -24.316037735849058
15.496567505720824  -24.787735849056602
15.503432494279176  -24.905660377358487
15.51029748283753  -25.200471698113205
15.517162471395881  -25.43632075471698
15.524027459954233  -25.613207547169814
15.530892448512585  -25.672169811320757
15.537757437070937  -25.613207547169814
15.54462242562929  -25.554245283018872
15.551487414187642  -25.43632075471698
15.558352402745996  -25.200471698113205
15.565217391304348  -24.905660377358487
15.5720823798627  -24.669811320754718
15.578947368421053  -24.316037735849058
15.585812356979405  -24.02122641509434
15.592677345537759  -23.785377358490564
15.59954233409611  -23.490566037735846
15.606407322654462  -23.254716981132077
15.613272311212814  -22.783018867924532
15.620137299771168  -22.31132075471698
15.62700228832952  -21.839622641509436
15.633867276887873  -21.60377358490566
15.640732265446225  -21.25
15.647597254004577  -20.448113207547166
15.654462242562929  -19.952830188679247
15.66132723112128  -19.245283018867923
15.668192219679634  -18.891509433962266
15.675057208237986  -18.419811320754718
15.68192219679634  -17.66509433962264
15.688787185354691  -16.88679245283019
15.695652173913043  -16.179245283018872
15.702517162471395  -15.471698113207548
15.709382151029748  -14.941037735849054
15.7162471395881  -14.64622641509434
15.723112128146454  -14.410377358490564
15.729977116704806  -14.174528301886792
15.736842105263158  -14.056603773584907
15.74370709382151  -13.997641509433961
15.750572082379863  -14.056603773584907
15.757437070938215  -14.174528301886792
15.764302059496568  -14.351415094339622
15.77116704805492  -14.528301886792452
15.778032036613272  -14.64622641509434
15.784897025171624  -14.82311320754717
15.791762013729977  -15.117924528301888
15.79862700228833  -15.35377358490566
15.805491990846683  -15.589622641509436
15.812356979405035  -15.707547169811324
15.819221967963387  -15.884433962264154
15.826086956521738  -16.06132075471698
15.83295194508009  -16.06132075471698
15.839816933638444  -16.179245283018872
15.846681922196796  -16.179245283018872
15.85354691075515  -16.120283018867923
15.860411899313501  -16.002358490566035
15.867276887871853  -16.002358490566035
15.874141876430206  -16.002358490566035
15.881006864988558  -16.002358490566035
15.887871853546912  -16.06132075471698
15.894736842105264  -16.120283018867923
15.901601830663616  -16.23820754716981
15.908466819221967  -16.474056603773583
15.915331807780321  -16.7688679245283
15.922196796338673  -16.945754716981128
15.929061784897026  -17.122641509433965
15.935926773455378  -17.299528301886795
15.94279176201373  -17.47641509433962
15.949656750572082  -17.653301886792452
15.956521739130434  -17.830188679245282
15.963386727688787  -17.889150943396224
15.97025171624714  -17.889150943396224
15.977116704805493  -17.889150943396224
15.983981693363845  -17.889150943396224
15.990846681922196  -17.77122641509434
15.997711670480548  -17.535377358490564
16.004576659038904  -17.358490566037734
16.011441647597252  -17.240566037735846
16.018306636155607  -17.004716981132077
16.02517162471396  -16.7688679245283
16.03203661327231  -16.41509433962264
16.038901601830663  -16.120283018867923
16.045766590389015  -15.943396226415093
16.05263157894737  -15.766509433962266
16.05949656750572  -15.471698113207548
16.080091533180777  -15.235849056603776
16.086956521739133  -15.058962264150942
16.09382151029748  -15.058962264150942
16.100686498855836  -15.058962264150942
16.107551487414188  -15.058962264150942
16.11441647597254  -15.0
16.12128146453089  -14.882075471698112
16.128146453089244  -14.82311320754717
16.1350114416476  -14.705188679245282
16.141876430205947  -14.46933962264151
16.148741418764303  -14.233490566037734
16.155606407322654  -14.056603773584907
16.162471395881006  -13.879716981132077
16.169336384439358  -13.584905660377359
16.17620137299771  -13.349056603773587
16.183066361556065  -12.877358490566039
16.189931350114417  -12.40566037735849
16.19679633867277  -11.816037735849054
16.20366132723112  -11.226415094339622
16.210526315789473  -10.518867924528301
16.217391304347828  -10.165094339622641
16.22425629290618  -9.81132075471698
16.23112128146453  -8.985849056603772
16.237986270022883  -8.13679245283019
16.244851258581235  -7.688679245283019
16.251716247139587  -7.452830188679243
16.25858123569794  -7.099056603773587
16.265446224256294  -6.509433962264151
16.272311212814646  -6.037735849056602
16.279176201372998  -5.742924528301888
16.28604118993135  -5.44811320754717
16.2929061784897  -5.212264150943398
16.299771167048057  -4.8584905660377355
16.306636155606405  -4.504716981132077
16.31350114416476  -4.033018867924529
16.320366132723112  -3.5613207547169825
16.327231121281464  -2.9716981132075464
16.334096109839816  -2.735849056603774
16.340961098398168  -2.382075471698114
16.347826086956523  -1.7924528301886795
16.354691075514875  -1.2028301886792452
16.361556064073227  -0.6132075471698109
16.36842105263158  -0.37735849056603854
16.37528604118993  -0.023584905660378297
16.382151029748286  0.4481132075471699
16.389016018306634  0.8018867924528301
16.39588100686499  0.919811320754718
16.40274599542334  1.0377358490566042
16.409610983981693  1.0966981132075464
16.416475972540045  1.0966981132075464
16.423340961098397  1.0966981132075464
16.430205949656752  1.0377358490566042
16.4370709382151  0.919811320754718
16.443935926773456  0.8018867924528301
16.450800915331808  0.625
16.45766590389016  0.4481132075471699
16.46453089244851  0.03537735849056567
16.471395881006863  -0.14150943396226445
16.47826086956522  -0.2594339622641506
16.48512585812357  -0.43632075471698073
16.491990846681922  -0.6721698113207548
16.498855835240274  -0.9080188679245271
16.505720823798626  -1.084905660377359
16.51258581235698  -1.2028301886792452
16.519450800915333  -1.3207547169811331
16.526315789473685  -1.4386792452830193
16.533180778032037  -1.5566037735849054
16.54004576659039  -1.6155660377358494
16.54691075514874  -1.6155660377358494
16.553775743707092  -1.6155660377358494
16.560640732265448  -1.6155660377358494
16.5675057208238  -1.5566037735849054
16.57437070938215  -1.3797169811320753
16.581235697940503  -1.1438679245283012
16.588100686498855  -0.9669811320754729
16.59496567505721  -0.790094339622641
16.60183066361556  -0.4952830188679247
16.608695652173914  -0.2594339622641506
16.615560640732266  0.09433962264150964
16.622425629290618  0.566037735849056
16.62929061784897  1.0377358490566042
16.63615560640732  1.2735849056603783
16.643020594965677  1.6273584905660385
16.64988558352403  2.158018867924528
16.65675057208238  3.514150943396227
16.663615560640732  3.632075471698113
16.670480549199084  4.4575471698113205
16.67734553775744  5.283018867924528
16.684210526315788  5.754716981132075
16.691075514874143  6.226415094339623
16.697940503432495  7.169811320754716
16.704805491990847  8.231132075471699
16.7116704805492  9.410377358490566
16.71853546910755  9.882075471698114


(As a reminder, SG1 in these graphs means Savitzky-Golay smoothing with a symmetric window of 51 samples and polynomial degree 1, and SG2 ditto with degree 2. SG1&SG2 means applying one after the other. The order doesn't matter. The acceleration graph above comes from applying central differences twice on the result.)
If I understand you correctly, you're saying your black acceleration curve was derived without using any Savitzky-Golay differentiation at all. IIUC, you're saying you smoothed the position data twice via conventional least squares, using best-fit linear for one smoothing and best-fit quadratic for the other. (Those two smoothings will effectively double the width of your smoothing window and will give more weight to central data points than to points near the edges of your extended window, so your process has at least that much in common with what I suspect was done by femr2.) Only then did you differentiate, using central differences instead of SG differentiation.

Please confirm or correct my impression.

In addition to the effects of SG smoothing, I wanted to mention that I have now looked at the Cam #3 data again, in relation to trying to determine whether the actual acceleration grew quickly or not. Cam #3 data is FAR less noisy, but it has issues with the unit conversion.

The following graph includes the velocity, as the derivative of the raw position data, and the acceleration. The vertical units are unspecified because I'm not sure what they are in femr2's data. Probably pixels. The horizontal units are frame number, which can be converted to time in seconds by dividing by 29.97. The "raw" velocity and acceleration use the axis on the left for units (albeit the axis is labeled units/frame, that's only for velocity; acceleration is in units/frame²). The "zoomed" acceleration's Y axis is the one to the right, also in units/frame². It has only one SG pass of degree 2 and with a 51 sample symmetric window.

[qimg]http://www.formauri.es/personal/pgimeno/xfiles/11-s/Cam3_profiles.png[/qimg]

But it suffices to look at the velocity plot to see visually how the average slope changes gradually, not abruptly, in the interval between frame ~270 and frame ~300 (modulo the bumps). That velocity plot is not smoothed, so that's a powerful argument in favor of a gradual onset of acceleration.
That's true, provided neither the positions nor velocity were smoothed in any way. Please explain the method you used to compute the velocity.
 
I'm inclined to think you're using a data set that's shifted relative to the data set femr2 and I used. Here's why I think so: My own curves were derived directly from femr2's data (listed below), independently of femr2's red curve, yet the noise peaks you see at the right side of my error graph line up vertically with the noise peaks in femr2's red curve.
Yes, but that's error in the raw displacement. Integrating a smoothed acceleration profile will give you a smoothed displacement. The noise peaks you see as matching all come from the non-smoothed (i.e. raw) displacement, that's why they match.


Here are the position data I've been using, extracted from femr2's DanRather1408.csv:
Same one I was using, yes. There are more decimals in my version but I doubt they will be noticeable at all.


For reference, here is my digitization of femr2's original "Savitzky-Golay" acceleration curve (the red curve):
Thanks. I wanted to check the error resulting from displacing that one by 0.1 seconds. I'll try when I can.


If I understand you correctly, you're saying your black acceleration curve was derived without using any Savitzky-Golay differentiation at all.
That's correct, as narrowed by the highlighted part (I did use SG smoothing). QtiPlot doesn't support SG differentiation. I can do it externally, but since that was a trial-and-error process that I performed while trying to find a match for femr2's curve, it was faster for me to try inside QtiPlot, which has a nice user interface that did that trial and error process progress more quickly.


IIUC, you're saying you smoothed the position data twice via conventional least squares, using best-fit linear for one smoothing and best-fit quadratic for the other.
If you mean the linear and quadratic fits that are part of the SG method, yes. Or I hope that that's what QtiPlot does, but I'm confident it is.


(Those two smoothings will effectively double the width of your smoothing window and will give more weight to central data points than to points near the edges of your extended window, so your process has at least that much in common with what I suspect was done by femr2.)
Let me emphasize that my explicit goal with that graph was to match femr2's as closely as possible.

I seem to remember that femr2 already said at some point that that graph used two passes of SG. My recollection might be wrong. I've found this graph that speaks about "one pass" and "two pass": http://femr2.ucoz.com/_ph/7/120219215.png


Only then did you differentiate, using central differences instead of SG differentiation.
That is correct (that's what I did), but it's also worth noting that I tried doing it in several different orders and I obtained the same result every time, no matter how much I zoomed. I guess there are mathematical reasons; I haven't stopped to think about it. And we already know that it doesn't matter in which order the convolutions (SG1 and SG2 in my case) are done, as in the end they are equivalent to the same single convolution.

Details on my experiments here: http://www.internationalskeptics.com/forums/showpost.php?p=8988689&postcount=2612

That's true, provided neither the positions nor velocity were smoothed in any way. Please explain the method you used to compute the velocity.
Regular central differences, which is the only method QtiPlot supports. I already went into full detail here:
I have also checked that the differentiation method used by QtiPlot consists exactly of averaging the slopes of the straight lines going from prior-to-current point and the current-to-next point. It should be obvious from the formula used in the source code, line 72 here: http://svn.berlios.de/viewvc/qtiplo...Differentiation.cpp?revision=1339&view=markup

[qimg]http://www.formauri.es/personal/pgimeno/temp/AverageOfSlopes.png[/qimg]

It's easy to prove that for equispaced data (hope that's a proper English word, the spell checker marks it wrong but "equiespaciado" is a valid Spanish word), which is the case for the video data, that formula's result matches exactly the slope of the straight line that goes from the previous point to the next point.
The link that points to the source code is now dead, unfortunately.
 
I'm inclined to think you're using a data set that's shifted relative to the data set femr2 and I used. Here's why I think so: My own curves were derived directly from femr2's data (listed below), independently of femr2's red curve, yet the noise peaks you see at the right side of my error graph line up vertically with the noise peaks in femr2's red curve.
Yes, but that's error in the raw displacement. Integrating a smoothed acceleration profile will give you a smoothed displacement. The noise peaks you see as matching all come from the non-smoothed (i.e. raw) displacement, that's why they match.
Good point, thank you. femr2 has mislabelled axes before, so that's a viable explanation.

Here are the position data I've been using, extracted from femr2's DanRather1408.csv:
Same one I was using, yes. There are more decimals in my version but I doubt they will be noticeable at all.
Thank you for confirming.

That's true, provided neither the positions nor velocity were smoothed in any way. Please explain the method you used to compute the velocity.
Regular central differences, which is the only method QtiPlot supports. I already went into full detail here:
I responded to that in the post immediately following.

I'm still unclear on whether you smoothed the position data before you computed the velocity.

I don't care whether the transition from rest was instantaneous (as Chandler claims) or took a full second (as femr2 has claimed), but I do have some interest in the mathematical methods that are available to answer such questions. My own implementation of Savitzky-Golay differentiation and smoothing is sufficiently modular to incorporate arbitrary weighting within the window, so I experimented with left-triangular and right-triangular weighting, comparing their results on the known stepwise acceleration to their results on femr2's position data for the northwest corner in the Dan Rather video.

I concluded the transition from rest to free fall was not quite instantaneous but did take less than half a second. I also concluded that NIST's estimate of 2.25 seconds of approximate free fall was about right, and might even be an underestimate for the northwest corner.

The graphs that led me to those conclusions are too busy to explain here. Once upon a time, I had intended to create simpler graphs for part 3, but gave up on that when I decided I was wasting too much time here.
 
I'm still unclear on whether you smoothed the position data before you computed the velocity.
No. You should be able to reproduce the same velocity curve by taking the raw data, discarding the endpoints and substituting each of the other points with the slope of the line that goes from the previous to the next point of the original data.

The raw data is FAR less noisy, that's why it looks so relatively smooth.

ETA: For further reference, this is the dataset used:
The vertical position data used (not plotted) is taken from http://femr2.ucoz.com/load/trace_data_nist_camera_3_raw/1-1-0-28, worksheet "NW Corner - 1.xls", first tab, column 52nd (extracted with xls2csv, so I don't know what exact column it is, maybe AZ, but I can say for sure it was labeled Y1).

ETA2: If you apply central differences as described a second time, you should obtain the red line in the graph, i.e. the one labeled "Acceleration (raw)". Note how the red and the orange line both start one sample after the black line, because of the discarding of the endpoints.
 
Last edited:
Thanks. I wanted to check the error resulting from displacing that one by 0.1 seconds. I'll try when I can.
I've done that now. By trial and error, I found that an offset of 0.11 looked more similar to my own than an offset of 0.1. Here's the result:

femr2_error_offset_0.11.png


EDIT: My own graph's error plot for comparison:
SG1SG2minusRaw.png


What was surprising for me is that the uncorrected plot:

femr2_error_uncorrected.png


is notably different from yours:

For the three acceleration curves I've been discussing here, that error analysis yields these results:

error.png
It is my understanding that it should match the red line, but it doesn't. What integration method did you use? Again, I've used qtiplot's. I will look it up later.
 
Last edited:
What was surprising for me is that the uncorrected plot:

femr2_error_uncorrected.png


is notably different from yours:

For the three acceleration curves I've been discussing here, that error analysis yields these results:

error20140813.png
It is my understanding that it should match the red line, but it doesn't. What integration method did you use? Again, I've used qtiplot's. I will look it up later.
Nice catch! As explained below, my plot goes wrong between 13.7s and 13.76s. (Because I intend to replace that plot with a corrected version, I've changed your link to make it go to a "permanent" copy of the current incorrect version.)

I used (my own implementation of) the composite Simpson rule, which assumes the data points are evenly spaced. I had thought my digitization of femr2's original "Savitzky-Golay" curve had that property, but there's a gap in the spacing between 13.7117s and 13.7666s. That accounts for the big upward spike in the error graph for both the femr2 and stepwise error curves (because both error curves used the same time points) near 13.75s. The error curve for true Savitzky-Golay doesn't have that spike because it was computed using times drawn from femr2's positional data.

When I fix this, the relative ranking of the error curves will probably stay the same but the error curves for both the stepwise acceleration and for femr2's "Savitzky-Golay" will improve so they aren't quite so much worse than the error for true Savitzky-Golay. If so, then the corrected graph will remain consistent with what I wrote in part 3 of my series on Savitzky-Golay smoothing and differentiation, and I'll just correct the graph on my web site (so the corrected graph will then appear within that post) while adding a URL to that graph that points to an explanation I'll post in this thread, citing your discovery of my mistake.

Thanks again for catching this. I may not get the corrected graph up until Monday, because I'll want to figure out why this happened when I digitized femr2's graph in early January 2013.

ETA: From looking at femr2's graph of his original "Savitzky-Golay" acceleration curve, his red line for that curve is hidden by a black curve within the interval where my digitization failed to recover data points for his red line.​
 
Last edited:
There seems to be something similar a bit after the 16 second mark.

As for the integration method used by QtiPlot, it claims to use Romberg's methodWP, taken from Numerical Recipes.
 
corrected error plot

In part 3 of my series on Savitzky-Golay smoothing and differentiation, the second graph uses double integration to show the error in three acceleration curves.

When pgimeno compared my original error graph to his similar graph, he found a couple of problems in mine. As I explained, these problems resulted from unequally spaced time points in my digitization of femr2's original "Savitzky-Golay" acceleration graph.

I have repaired those problems by digitizing one of femr2's cleaner presentations of that graph, interpolating two points by hand where his cyan line obscures the red "SG" curve. I then recalculated the error plot, using the same file name so the corrected graph will show up when readers view the part 3 I posted several days ago. I also placed a URL within that corrected graph that points to this explanation.
 
“Holy Zombie Thread, Caped Crusader.!!”

I see that this forum is about to be closed down shortly, so I thought that I’d toss this one out here before it left.

About a year ago, I was walking on the beach & had a revelation about one of femr’s claims regarding “precision of his measurements”.

Moments of quiet are good for revelations.

I’d objected several times that he’d never provided any proof of his 0.1 pixel, or .01 pixel, or .001 pixel claimed accuracy. He did a sloppy, imprecise, undefined bit of hand-waving using a black circle on a white background, that had no application to the images that I could see. And, as usual, he was unable to provide any mathematical justification for his claims. And petulantly refused to engage in any real discussion.

I knew that his argument was BS, but wasn’t interested in going into the proofs. And he didn’t provide sufficient info to get into the details. I decided that it was HIS job to prove is claimed accuracy, not mine to prove his claims wrong. As usual, it remained an unproven assertion.

But I stumbled on a very elegant proof that his accuracy could not possibly be better than about ±0.5 vertical pixels. The elegant part is, one doesn’t need to know any of the details in order to prove it.

Here it is:

Each frame of the interlaced video (from US NTSC broadcast & video tapes) consist of an upper (U) & lower (L) field.

In analyzing the fall of WTC7, femr decided to use only one field of each frame (let’s say U, it doesn’t matter which one) of one of the videos. He felt that this provided him better … something.
Accuracy? Image registration? Lack of timing adjustments? (I don’t recall.)

Then he mashed together the single fields (eliminating every other, lower frame, line) & claimed that his tracking points traced the fall of the roofline within 0.01 or 0.001 (or some absurd) sub-pixel accuracy, within those remaining fields.

Here is the problem: the location of the point that he was tracking could only be located within the lines of that upper field about 50% of the frames, or 50% of the indexed time.

The mashed-together lines of the single-field image was NO LONGER an accurate representation of the image of the building. In order to accurately represent reality, each horizontal line of his image needed to be separated by an equal width blank line, representing the unused field.

Since the actual location of the roofline point was accelerating, then pure probability REQUIRES that the real location be located on some line in the U field ~50% of the time, and on some line in the L field ~50% of the time. In order to be accurate, around half the time, his software should have returned an answer “somewhere between this line & the next one, but I can’t figure out where.” There is no way around this.

However, femr’s calculation never returned this statement. His method returned a roofline location at every point in time. His method FORCED the roofline to be located ONLY within some line of the single field that he used, EVERY TIME.

The two paragraphs above are mutually exclusive.
The answer is obvious: He is off by about ±0.5 vertical pixels on at least half of his data points.
The process also throws into question the calculated location of the roofline for ALL points in time, since the scaling factor becomes so non-linear.

QED.

Replies welcome.
Snark … not so much.
 
Last edited:
I thought that he used both fields in succession, "de-interlacing" the frames. He had a chart labelled "Dan Rather - 59.94fps" IIRC.
 
Since the actual location of the roofline point was accelerating, then pure probability REQUIRES that the real location be located on some line in the U field ~50% of the time, and on some line in the L field ~50% of the time. In order to be accurate, around half the time, his software should have returned an answer “somewhere between this line & the next one, but I can’t figure out where.” There is no way around this.
It's not correct that there's no way around this. There are two primary ways around this.

First, the software tracks points within regions, not in isolation, and matches the full region originally tracked against the data of every frame, which enables it to derive the new position of the tracked point from the data of the full region tracked.

Just as an illustrative example, even if it was not the case for the corner, imagine there are a few windows in the tracked region. The borders of these windows would disappear and reappear as the building descends, and by pure probability (unless the separation of the windows is too close to a multiple of that of the video lines by chance), it's unlikely that they do that all at the same time, so the software does have more information about the position of the whole block based on them, and can thus derive the position of the tracked point with more accuracy. It does not need to "see" the tracked point all the time, which is where your assumption goes wrong. It just needs to determine the overall position of the tracked block, in order to derive the position of the tracked point from it.

That would be the case for the static point in the Dan Rather video. Here's the frame that femr2 posted in page 1 of this thread:

759212128.jpg


However, that argument doesn't apply to the region tracked in the Dan Rather video. That's where the second argument comes.

The second way around it is the fact that the Dan Rather video in particular was blurry enough for the border to span several lines of each field. Here's a zoom over the tracked portion of the still interlaced image, so you can see what I mean (the JPG above is too low quality even for demonstrative purposes, so I grabbed a video from the NIST database and extracted a frame):

DanRatherCornerZoomed.png


Due to whatever process produced the smearing (be it digital zoom, poor focus, analog effects, or whatever), the border spans about 6 pixels vertically for the traced corner. This implies that the software has not just one fraction of a field's line, as you seem to imply, but about three lines for each field from which to gather intensity information that helps it determine the position of the traced block. (Paradoxically, if the image was more out of focus, the position would have been more precisely determined.)

Apart from that, I don't know if that's the case, but I think it's plausible that the software has an interlaced video option to account for the non-linearity of the position between lines of each field.

So, for the reasons above, the paragraph I've quoted is not right, and thus your conclusion doesn't follow.

ETA: Edited picture to add a marking for the 6 pixels.
 
Last edited:

ISF - Join now!

Every member here is approved by hand. No bots, no spam, just people who care about evidence and honest debate.

Membership is free!

Create your free account

Back
Top Bottom