Register to reply 
The Monte Carlo Simulation 
Share this thread: 
#1
Apr614, 03:06 AM

P: 10

The Monte Carlo simulation is a very important tool in particle physics specially to tune in the preciseness of the real time experiment. In particle physics, I had an opportunity to work on the data analysis of neutrino flux produced from the g4numi (the Neutrino beam from main injector, in FERMI lab). The simulation gives information like the production vertex of parent neutrinos, their velocity, momentum, the production vertex of neutrinos and other important information besides the regular neutrino flux obtained at different energy spectrum. I was mainly curious about the monte carlo itself. I tried to find a simple explanation on the design of the monte carlo like how do they determine the geometry of the beamline, the particle trajectory at various places (like the target where the proton hits, the decay area where mesons decay) and so on. How do they feed the probabilistic paths and particle interactions that take place in real life and how do they design or control the big events like hadron showers in special cases?
I am not sure if it was a proper place to pose this question but any guidance will be highly appreciated. Thank you. 


#2
Apr614, 03:05 PM

Sci Advisor
P: 6,040

I am not familiar with your particular case, but I can try to answer it in general terms. The physics theory (including experimental observation) gives a mathematical model for the probability distribution of the process being described.
For example the angular distribution of neutrinos resulting from a given reaction(direction and azimuth). The Monte Carlo process starts with two random numbers (uniformly distributed between 0 and 1) and selects two angles using the given distributions. http://en.wikipedia.org/wiki/Monte_Carlo_method 


#3
Apr714, 12:58 AM

P: 355

Well, there are various pieces that go into the Monte Carlo. There are codes that deal with each of the bits. You basically have to simulate the hard partonparton scattering event, taking into account the beam energy/geometry, which generates your new particles; do the probabilistic decay of these things, and simulate the transport of this radiation/matter through the detector, and model all the interactions with the detector material as well as the detector response (i.e. the electronic readouts). The vast majority of CPU time is burned on this detector simulation. ATLAS and CMS have full 3D models of their entire detectors, down to quite small details, built using a package called GEANT4 (http://en.wikipedia.org/wiki/Geant4), and this full simulation is run for thousands/millions of collision events. Again, the CPU requirements are immense.



#4
Apr814, 02:33 AM

P: 83

The Monte Carlo Simulation
The main benefits are the stochastic uncertainties.
For a nparticle final state, each with different momentum configurations. That's a lot of degrees of freedom. So other types of simulation fail. The use of pseudo random numbers in the generation also mimics the quantum probability. You don't know exactly what an event will be before you measure it... Also, as people have mentioned, the full simulation generally has several sub processes. These are implemented sequentially by MC methods. So you can chain up several complicated independent sub processes quite easily. Hope this helps 


Register to reply 
Related Discussions  
Monte Carlo Simulation  Set Theory, Logic, Probability, Statistics  9  
Monte Carlo simulation  Atomic, Solid State, Comp. Physics  0  
Latticegas Monte Carlo Simulation  Classical Physics  2  
MonteCarlo simulation for a matrix model  General Physics  1  
Markov Chain/ Monte Carlo Simulation  Set Theory, Logic, Probability, Statistics  0 