Physics Of 9/11 Events - Part 3, continued from: 9/11 Events- New thread

really?
what about all those terrorist watch lists? they ought to be good for stopping terrorists.
or enabling them.

Only luck or 20/20 hind-sight could have stopped the jet that hit the Pentagon.

There are always armed jets just outside of DC, waiting, just in case.

They had enough time to scamble a man on a bike with a shoulder launched missile.

Total Horse Puckey.


All of it is true. You are the one sitting in a pile of poo...

Really?

Source from 2001 showing this assertion is TRUE please.

Arthur

Here are two hypotheses which cannot be distinguished by the naive Bayes factor method (C447 data):

Both are vertical avalanche style, the difference being the mass used in the avalanche part of the resistive force function; the best used just M_B, the mass of the crushed materials; the other uses the total moving mass (M_A+M_B).

Here, by naive Bayses factor method (C447 data) the poorer hypothesis is right on the edge of being substantially disconfirmed. But other statistics (not shown) show that the curve shape is poor, which tends to help disconfirm it.

Edited to add: WIth quite different parameters, the poorer hypothesis is 3.6 dB down on the C449 data. This parametric instability is certainly a mark against the hypothesis.

This post has been edited by David B. Benson on Jan 13 2008, 10:59 PM

Reducing the effective mass makes a better fit? Interesting. I'd be curious as to fitness as a function of mass for that hypothesis. Any other values of mass you've run on these?

DBB,

I never replied this one. This originates about 10 pixels below the impact zone, or about 4-6 meter. The object is about 3-7 meter and the initial horizontal speed is about 13.6 m/s, I didn't calculate the real error margin but I guess > 10 m/s
The tower wind cannot account for that. Released strain energy or a kind of scattering effect or wedge effect sounds more plausible. Tiddlywink effect....?

There was someone who mention the triplet structure of the perimeter columns, that is indeed a possible debunk, but i'm not yet convinced, because that still doesn't explain why that set of dust ejections is also a row at the right hand side. If I can find the time and the videos I want to synchronize videos at all 4 sides. I also would like to see how fast that collapse really was. That's also very useful for David's fitting function I guess.

Only slightly better. The reasoning is that the crushed materials are continually re-crushed resulting in the term kM_Bv^2. I have also tried kv^2 (for crushing only at the crushing front) which is even worse.

Here is the vertical avalanche resisting force applied to C447 with a constant stretch of 0.18:

Substantially disconfirmed, but ok with a larger stretch, but not parametrically stable.

I opine it was a piece of aluminum cladding, somehow tiddlywinked out...

Parametric stability, intuitively, seems a good indicator of fitness of model. Is there any formal argument in support of the notion?

Somewhat related:

I've recently played around with conventionally fitting a known generated dataset injected with known error (so far just time and value offsets). The sensitivity of fit is significant, particularly wrt time as there is an asymmetry about t0, where an analytic representation of a curve diverges wildy from the graph it attempts to fit.

Trimming the 'constant' baseline at the beginning of a dataset is obviously not the same as t0 determination, which is crucial, and I appreciate that your method attempts to minimize the variance in the high grade hypotheses wrt time offset.

I screwed around with real numbers and a curve fitting app rather than try to develop any sort of formal analysis to generalize the work. What I found in checking polynomial fits might be old news to some and not even surprising to me, but two things were most interesting:

1) the asymmetry of the graph against the time coordinate made the fit far less sensitive to time offset error in one direction versus the other
2) higher order polynomials often provided the best fit for even slight time errors, despite the fact that test data was generated with certain coefficients identically zero

Intuitively, for the limited case I examined, I'm inclined to cast similar suspicions on any terms with small coefficients. However, my broader common sense tells me that this is an invalid operating principle, as the reverse is probably also true: simpler forms can fit better than the true, underlying complex form, if the influences of the terms are but subtle.

Edit: I did not state #2 correctly. What I meant was non-zero coefficients provided a better fit than did the real form with no term at that order. I do want to check the fit of higher order unconstrained solutions versus lower order constrained.

This post has been edited by OneWhiteEye on Jan 14 2008, 08:49 PM

Could not have termed it better, good turn of phrase. It doesn't have to be a wedge effect as you graphically depicted, it can be pure strain energy.

In looking at the north tower video we've kicked around, I swear I see obvious deformation of the 'undamaged' lower portion in response to the increasing eccentricity of the load from above. Unable to quantify yet - I've tried, and to this point, I don't even have a way of making it visually obvious, though applying translucent regular patterns to produce artificial moire seems promising.

Even large components could be launched by the tiddlywink effect. I just don't think a significant amount of material can go that route. Tiddlywinks: small mass, large strain, externally imposed.

My grandma had the "vlooienspel" at her home, flea game if you translate it literally but glad I got a dictionary, I still need it every day!

In that controlled demolition video from national geographic where they placed charges on two levels also such a piece pops out, but then caused by explosives. During the rest of the drop the ejections don't look very violently. It mainly stays within the footprint.

I've looked for 'parametric sensitivity'. Not the right notion.

What I call parameter estimation is called 'parameter calibration' by some. That didn't help.

I'll try some more and then likely have to devise a suitable argument.

However, the intuition is clear: adjacent pixel columns describe essentially the same phenomenon. If noticeably different parameter values are required for these by some model, then no matter how appealing the physics underlying the model, it is simply to sensitive to provide stable predictions.

This post has been edited by David B. Benson on Jan 14 2008, 10:43 PM

Thank you for making the effort. Vlooienspel would not have made the same impression.

A certain JPL scientist we know (not Jack Parsons) has made some pronouncements on the abilities of explosives to propel structural elements. I haven't looked at it, but I wonder how well it jibes with a structure that is known to have members propelled by explosive.

Is the piece you refer to ejected too early to be due to anything other than the charge(s)?