PDA

View Full Version : Monte Carlo Visual Basic Simulation




Max Payne
Mar 5, 2007, 03:11 PM
Hello all MR members...

I have been doing experimental work in my engineering study. Now, I have been asked by my supervisor to build a monte-carlo simulation program. The program will use some of the experimental results to predict the rest of the results. Something like Artificial Neural Network. How can I make it simpler?

This is what I am planning to do.

1. Feed some of my results (i.e. 60% of my results) in the program.
2. The program will predict the rest of the results (i.e. 40%).
3. Compare the simulated results with the experimental results and calculate the deviation.

Do you guys think this is possible? I have worked briefly with C programming & VB, but I am no expert. The Monte Carlo program I did previously was to predict the probability of something by using an equation. Which is a bit different than what I am planning to do now.

Thank you all and I appreciate your help.



robbieduncan
Mar 5, 2007, 03:23 PM
I'd not use VB. I'd use Java, C# then C++ in that order of preference. VB (unless you are talking about VB.Net) is just not going to cut it.

Max Payne
Mar 5, 2007, 03:43 PM
I'd not use VB. I'd use Java, C# then C++ in that order of preference. VB (unless you are talking about VB.Net) is just not going to cut it.

Thank you robbieduncan. I used Excel VB previously when I worked with Monte Carlo, but the simulation was very simple. I have no background on java or C#. I self learned C++ and I studied a module in C. I think I can handle C++ or C# if I'll have to use it. I wonder if you or any other member can help me, if what I am planning to do is possible?

robbieduncan
Mar 5, 2007, 03:57 PM
Pretty much everything is possible, although with your experience it'll be amazingly challenging. This is not the sort of thing to be taken on lightly.

Monte Carlo methods are not something I have any personal experience of. I did study Neural Networks back at Uni and would say they are very different!

Max Payne
Mar 5, 2007, 04:27 PM
Pretty much everything is possible, although with your experience it'll be amazingly challenging. This is not the sort of thing to be taken on lightly.

Monte Carlo methods are not something I have any personal experience of. I did study Neural Networks back at Uni and would say they are very different!

I'm planning to do Monte Carlo because I want to run away from Neural Networks :D

lazydog
Mar 5, 2007, 05:20 PM
Hi,

Go for it! Once you have your algorithm and maths worked out writing the code to implement it won't be too bad I think. At least you will know what you have to do and so with a little help from friends I'm sure you will be able to do a good job. Stick to writing a command line program, ie no fancy user interface etc. Just dump your results out to a file and use a graphing application to produce the nice output. You'll get lots of help implementing code from forums if you can be specific in describing your coding problems. Unless you need to use specific libraries I don't think it will make much difference to the end result what language you use. However if you don't know and don't want to learn object orientated programming just for this project then stick to C I think.

hope this helps and good luck!

b e n