monte carlo simulations

Hi everyone ,

what is Monte Carlo simulation ?

i can find this term in : management ( risk analysis ) , RF planning ( traffic calculation ) , etc...

Best regards

Monte Carlo simulation is a simulation in which the circuit components will have tolerances (5 to 10%) and number of runs(simulations) will be done and outputs for

each run will be collectively plotted .

It is also known as worst case simulation.

Monte carlo simulation is a statistic method not only related to electronics, but more in general is used to investigate phenomena in which there are many random variables varying following a known distribution (uniform, gaussian, etc.), like, in electronic the tolerance of components, substrate thickness variability, etc.

As said by Prashanth.vinnakota, the simulator will generate many runs, each having a different value for the set of random variable (however inside a defined range of variation). Each set will be generated so that all the variable will be uncorrelated each other (usually a chi-squared test is used to check the condition). At the end of the simulation the result of all the runs is presented in text or graphical format.
The results will represent the behaviour of the system taking into account the variability of the parameters.

If a large enough number of runs is launched then the worst case (or almost worst case) can be identified.

Hi albbg ,

to investigate phenomena in which there are many random variables varying following a known distribution (uniform, gaussian, etc.)

Click to expand...

Does it mean when i want to do a simulation i can choose the distribution that i want to the specific variable or there is a condition to choose such distribution ?

(usually a chi-squared test is used to check the condition)

Click to expand...

Could you please elaborate more on this test ?

Monte Carlo is now also a way of perceiving history, as suggested by Taleb in Fooled by Randomness. Interesting and fun book, I'm reading it now.

Usually it is possible to choose the distribution that fit with that you expect to have. For instance the Microcap circuit simulator allows a choice between gaussian and uniform distribution.
The chi-squared (or chi-square, I'm not sure) test is used during the random number generation to chek their independency. Please have a look to:

https://changingminds.org/explanations/research/analysis/chi-square.htm