I'm looking for suggestions on books/articles
that explain monte carlo/ quasi monte carlo
techniques. In particular on how to interpret
the results. I would even be interested in
BJ algorithms, but I have in mind a different project.
Any suggestions would be appreciated.
> I'm looking for suggestions on
> books/articles
> that explain monte carlo/ quasi monte carlo
> techniques. In particular on how to
> interpret
> the results. I would even be interested in
> BJ algorithms, but I have in mind a
> different project.
> Any suggestions would be appreciated.
What, specifically, are you looking for? The general principle behind MC techniques is simply to feed a computer as much data about a particular "event" (sports, finance, etc.) as possible, and then have it spit out scenarios, on a repeated basis. Then, depending on frequency of results, you ascribe a probability to each outcome, based on its relative frequency.
Don
> What, specifically, are you looking for? The
> general principle behind MC techniques is
> simply to feed a computer as much data about
> a particular "event" (sports,
> finance, etc.) as possible, and then have it
> spit out scenarios, on a repeated basis.
> Then, depending on frequency of results, you
> ascribe a probability to each outcome, based
> on its relative frequency.
> Don
Rookie question, perhaps. How is this different than a back-propagating neural net?
> What, specifically, are you looking for? The
> general principle behind MC techniques is
> simply to feed a computer as much data about
> a particular "event" (sports,
> finance, etc.) as possible, and then have it
> spit out scenarios, on a repeated basis.
> Then, depending on frequency of results, you
> ascribe a probability to each outcome, based
> on its relative frequency.
> Don
Yes, now that I have thought more about it, I
see that is all that there is to it. I was a little concerned about how to calculate the
errror in estimating the true probabilites
from what is collected by the random sampling,
but I think I see it now. Its been almost 30
years since I 've done this sort of thing.
ComboProf
> Rookie question, perhaps. How is this
> different than a back-propagating neural
> net?
Monte Carlo techniques do not involve feedback or learning. If you determine the probability of heads by flipping a coin 10,000 times and counting how many times heads appears, that's Monte Carlo.
Below is a link to an applet that uses a Monte Carlo technique to approximate the value of Pi.
Posting Permissions
-*-
Bookmarks