What's a Monte Carlo Algorithm?

[JS80]

So I read Wiki and I don't really understand it. So is it suppose to be like a fortune teller? Can someone explain this in uber layman's term?

 

[DVK916]

Is that the same as Markov Chain Monte Carlo?

It has to do with probablility. Statistcians use it.

 

[EvilYoda]

It's just a method for simulating a scenario

 

[illusion88]

Monte Carlo makes a good cigar

 

[ijester]

It is a method used to study something when you need a 'Semi-Random' starting point to effectively determine behaviour.

Something like a roulette spin, or dealing cards from a shuffled deck. Both are random, but within set limits.

 

[IronWing]

Monte Claro simulations are brute force method of simulating physical systems for which you have imperfect knowledge. One example might be the movement of pollutants in groundwater. Say you have a few soil samples that tell you a bit about the characteristics of the soil but you don't know the exact size, shape, or mineralogy of every single soil grain in your aquifer. You do however have enough samples to say something about the distributions of size, shape, and mineralogy of the soils in your aquifer. You also have well points telling you the hydraulic head at a few points, but a not a perfect map of hydraulic head. So you set up a computer model of the groundwater flow/pollutant transport through the aquifer. Traditionally you would evaluate your data and and estimate an appropriate distribution of hydraulic head and soil properties in ythe aquifer and run the model and see if the output reasonably matchs any data you have about pollution distribution. With a Monte Carlo simulation you would run the model many, many times, varying the input parameters using random samples shaped by the known distributions of the input parameters. You would then gather the results of the multiple runs to come up with some type of "mean" solution and an estimate of the "standard error" about the solution.