I am simulating a bilayer and intend to calculate some dynamical properties. I tried to calculate the relaxation time for the chains in the hydrophobic region. For this I choose a vector along the chain and calculated the angle of each chain vector from every frame. I used this angles as an input to calculate the relaxation time using FORTRAN code from this website: - http://math4u.wordpress.com/2009/09/30/autocorrelation-in-fortran/. (I have modified this code according to my system).
I have shown the plot I get in this attachment (please view).
Related to this I have two questions which I am seeking for answers or explanations.
First, I get negative values at larger times when calculating auto correlation function. Is this acceptable for bilayer chains in L-alpha phase (liquid crystal phase)?
Second, If I want to calculate the relaxation time I need to calculate the area under the graph. For that I need to fit my data to a single exponential function. Since I have negative values, how can I fit the data using exponential function, since this function takes care only positive values ?
Hello.
First of all, what is the software you are using? If it's GROMACS, you hardly need that king of calculations. For instance, you have:
-> g_density: for calculating membrane density;
-> g_msd: when calculating lateral diffusion;
-> g_order: deuterium parameters for hydrophobic tails (I think that is what you need).
For lipid areas, you can get a small perl script (GridMAT_MD.pl) from http://www.bevanlab.biochem.vt.edu/GridMAT-MD .
Best regards,
Ricardo
there is nothing wrong in getting a negative values, as you can get anti correlated at some chosen lagtime. Ideally for long enough sampling, it should start to oscillate.
However, if i understand right you are using angles which are circular variables and if you calculate linear correlation with them you will get spurious results.
- Badges
- Science topic
- Similar topics
- Chemistry
- Theoretical Chemistry
- Computational Chemistry
- Molecular Dynamics