[lug] Perl, sorting hashtables by value, and floating-point

[D. Stimits]

...
> Curious. I've read a little about Monte Carlo methods
...

A short blurb on Monte Carlo, which is used extensively in finance. It 
is like a coin toss experiment. If you measure enough coin tosses, the 
curve generated happens to be a normal curve. If you have a few coin 
tosses but not a lot, it is the student's t-distribution. If your coin 
is weighted non-50:50, then you will get skewness and kurtosis.

In finance there are plenty of experiments (especially in bonds, 
options, and futures) where you can't measure how much a security will 
change by, but you can predict over a very short term the odds that it 
will go up or down...a coin toss...where the weight of the odds are what 
are put in from market information. Then by creating a large number of 
coin tosses and accumulating them over time, you begin to get a curve of 
probabilities for longer time periods. If you have a situation in 
linguistics that you can decide a yes/no on it, and assign weights as in 
a fair coin or a coin that is weighted towards heads or tails, then you 
can do the same thing in linguistics. [look up the binomial theorem if 
interested, that's what a coin toss is]

D. Stimits, stimits AT comcast DOT net