URGENT: Monte Carlo

I need info about Monte Carlo method used for determination of measurement uncertainty. (procedure is to be done in Mathcad) Maybe some books (free to download) to recommend or sites? Or just site with free engineering books? Anyone? Anything? I'd be very grateful.. Thx a lot!

Some of that might be found at

.

DD

Maybe you will have luck, when contacting someone in the QMC@HOME Forum. They do Quantum Monte Carlo research, using volunteered CPU's. I do not know if they use Mathcad or not. (I participate there as volunteer too, so you could refer to me.)

Good luck

EXCEL has Monte-Carlo capability. Works fine.

OOps

There was an ASME paper last year on this subject (applied to gas turbine performance testing) earlier this year: Spencer, Friedman, Sullivan; Combined Cycle Performance Test Uncertainty Validation by Comparing Monte Carlo Analysis with Monovariate Perturbation Results, Proceedings of Power 2006, May 2006.

Google also turned up the following papers:

I not aware of a good website that describes the fundamentals of Monte Carlo analysis applied to measurement uncertainty. My basic understanding is that you perform a large number of simulated "experiments" with the inputs randomly varied and review the variation in the results. The amount of variation (and possibly the distribution) of the input variable must be specified.

David Parker West Palm Beach, FL

djo wrote:

Although this website is for monte carlo simulation using excel, you may be able to download the spreadsheet and examine the formulas to help design your algorithm. The text may also be of use.

also check out the wiki:

Here is what little I know about it:

Suppose that you measure two quantities, x and y. From these two numbers you calculate a third number z, which is a function of the two measured quantities x and y, i.e., z=F(x,y).

This is the problem of error (or uncertainty) propagation. Presumably x and y are random numbers, and you know what the uncertainties are for x and y, and you want to find the resulting uncertainty in z.

If F is a fairly simple function, the usual method is to take partial derivatives of F with respect to x and y, and use the standard error propagation equation. See:

This requires that: 1. You can actually take the required partial derivatives of F. 2. That the errors in x and y are small, so that the function F can be approximated by a first-order Taylor series expansion. This approach also doesn't tell you anything about the distribution of the random variable z.

The Monte Carlo method is an alternative. You know the distributions of the random variables x and y, so you use an random number generator to generate a large number of (x,y) pairs. You take each (x,y) pair and plug it into your function F, and generate values of z. You do this many times. You collect many values of z. From these you can find the standard deviation of z, or plot a histogram of z, or whatever you like.

Olin Perry Norton

Hey Doc!

Could "djo" actually be Dennis J. Ols>>I need info about Monte Carlo method used for determination of measurement

The world is full of possibilities, aye.

DD