Tuesday, February 17, 2009

Update: Reply from Hans Kristian

I wrote the authors of the Bolt paper. Here's the reply




Dear Mark,

First, thank you for your interest in our paper! It is both interesting
and fun to see how widespread interest this paper has received :-)

> Dear Professors Eriksen, Kristiansen, Langangen, and Wehus,

(I should say that none of us are professors, actually -- we're just
four post-docs having a bit fun ;-))

> I've just read your article on Usain Bolt's 100m world record
> published in the American Journal of Physics. As a track fan, I
> jumped right to your article when I read the March table of contents,
> and I was intrigued by your analysis. I thought your method of
> estimating the position of the runners was quite ingenious, and was
> impressed by how much you got out of the freely-available broadcast
> and internet videos.

Thank you! The hard part, as you say, was definitely to extract the
measurements from the available videos. Fortunately, both the NBC and
NRK footage were of quite good quality, and with some care (and several
cross-checks) it was in fact possible to get reasonable measurements
from these :-)

> I had two questions regarding the paper that I hope you might shed
> some light on. First, I was curious about the method of estimating
> the uncertainty in the runners' positions. From the graphs of
> Thompson's and Bolt's positions, it seems likely that most of the
> uncertainty is systematic (common).

That's true.

> But I was wondering what the procedure was to figure out how much the
> position estimates might be off.

To be honest, this was done by a fair amount of "guestimation". For each
screenshot, we tried to draw a best-fit line orthogonal to the track,
using various known and fixed points in the track. So this typically
amounted to making various triangulations througout the track. However,
once we had decided on a given best-fit line, we also drew a few other
lines around this, to see how much "slack" there might be, and these
were the basis of the uncertainty estimates. It was difficult to be
really quantitative about this, but in most frames, I don't think we
were too far off either -- it was usually quite straightforward to see
if a line made sense or not.

> For example, were you estimating the error in the slope of the lines
> across the track used to calculate the vanishing point, or estimating
> the error in the positions of the tops of the runners' torsos relative
> to the projected lines across the track (which was almost 2m below
> Bolt's shoulders), or using some other method?

As far as the torso goes, the first step was to mark a point just below
the runners chest in the track. Then that was used a pivot point when
drawing the lines. Of course, there was some uncertainty in marking
these points as well, but I think the angle was the dominant source of
error.

> Also, I was curious as to why the analysis included only Thompson and
> Bolt. It might be interesting to see how robust the results are with
> respect to the runner used for calibration purposes. Maybe selecting
> the runner closest to the rail would yield the least random
> uncertainty because he is closest to the reference bolts? Or maybe
> Bolt's acceleration profile could be compared to the average of the
> field?

These are definitely good ideas. The choice of using Thompson as a
reference was quite arbitrary, and essentially only given by the fact
that he came second. But, as you say, it might have made more sense to
use one near the rail, especially since this probably would have made
the two profiles more "orthogonal" in terms of errors -- the systematic
errors might be smaller when measuring distances for two runners far
away from each other..


However, a main point in this analysis is that it is a fun example of
simple physics, and not really a rigorous kinematic analysis of Bolt's
performance. So we decided to wrap it up with the present paper, and
leave it at that :-)

Again, thanks for your interest and suggestions! :-)

Hans Kristian

1 comment:

Anonymous said...

Men and women body builders are always searching for the perfect community to encounter. Numerous have never reasoned being online with others who hold large knowledge about which workout vitamins and proteins to accept in order to grow with results. This is because weightlifting is a sport, and with any sport you go where the fanatics are to learn about it! To read more about Female Weight Training

To press the statement even more, you may be able to go and learn from a individual trainer about precise techniques but their noesis will be only be so overwhelming and wide. In order to find out you must go where other like you talk together to study about one topic. This is cause a single bad lift or protein can break everything you possess, worked so long for already. Still more so, there are few groups
and places you can go in order to conversate about steroids and additional drugs used in lifting.

But even if lifting weights and gaining muscle isnt what you are digging for, many Weight Loss Programs are around solely for individuals to discover how to lose those added 5 pounds, and with the society of today it seems everyone is getting larger and fatter. So no concern who you are in the end, holding a firm diet and the correct exercises can create all the difference!