I'm working with interaction between peptide-lipid bilayer systems, and I was wondering how to correctly analyse and show the statistics of my simulations. I mean, are there one correct way to express some results with average values and error? Does anyone know one paper or other source where can I find and learn about statistics in MD simulations?
Thanks in advance!
I can recommend good (although slightly older review) concerning statistics and uncertainty in MD:
http://www.sciencedirect.com/science/article/pii/S1574140009005027
Hello Danilo,
in Monte Carlo and Brownian dynamics it's very important to check if the measure that you are taking are correlated. Imagine that your simulation lasted for a time "x" and during this time, by chance, the interaction that you were observing was acting. If you use the normal average and standard deviation you will get a value. Now imagine to run the simulation 10 time longer and observe that only in rare case you have an interaction, and usually this interaction last for "x" time. Here the measure are correlated, and the simulation, to give a correct result for the average and the error should last much longer than "x".
What a person can do is to look at the autocorrelation function of a given observable and use the blocking method to estimate the error.
Here there is an introduction to these technique:
www.hri.res.in/~prasen/MC-erroranal.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics