[gmx-users] time-block analysis

Christopher Neale chris.neale at mail.utoronto.ca
Tue Oct 30 02:03:11 CET 2012


Dear Leila:

I presume that "time-block analysis" is the same as block averaging, but it is impossible to be sure without more 
context. I suggest that you find the person who suggested that you do "time-block analysis" and ask them.

Very generally, you will need to do your own scripting to do block averaging. Look at Allen and Tildesley for
a good description of block averaging or look at the original (as far as I know) reference:
Flyvbjerg, H.; Petersen, H. G., Error estimates on averages of correlated data. J. Chem. Phys. 1989, 91 (1), 461-466

Basically, you will need to take your raw data and make block averages. Here is a very simple script to
do that for a 2-column data file that has time in the first column and some other data in the second column

MIN=0
MAX=1000
BLOCK=100
for ((min=MIN; min+=BLOCK; min<MAX)); do
  let "max=$min+$BLOCK"
  cat data.file | awk '{if($1>='${min}'&&$1<'${max}'){s+=$2;n++}}END{print ('${max}'-'${min}')/2,s/n}'
done > my.block.averaged.data

I am sure that there are more elegant awk scripts that can do this, but the intention here was to show how it is 
done most clearly.

Chris.

-- original message --


Can I do time-block analysis by gromacs? If so, which command?

Is time-block analysis the same as block averaging?


Any help will highly appreciated.



More information about the gromacs.org_gmx-users mailing list