[gmx-users] Re: REMD average acceptance ratio

[bharat gupta]

Dear gmx members,

I have posted the same question previously , but I didn't get any reply.
So, if anyone can help me out ...

I performed a REMD simulation on a peptide 384 atoms (24 residues). In
total 11 replicas were simulated for a period of 50ns each. The exchange
was allwoed at every 1000 steps. The output of md.log file is :

Replica exchange statistics
Repl  24999 attempts, 12500 odd, 12499 even
Repl  average probabilities:
Repl     0    1    2    3    4    5    6    7    8    9   10
Repl      .16  .16  .16  .17  .18  .21  .24  .26  .28  .30
Repl  number of exchanges:
Repl     0    1    2    3    4    5    6    7    8    9   10
Repl     2038 2007 2065 2117 2182 2587 3022 3213 3554 3703
Repl  average number of exchanges:
Repl     0    1    2    3    4    5    6    7    8    9   10
Repl      .16  .16  .17  .17  .17  .21  .24  .26  .28  .30

The acceptance ratio for each replica and average acceptance ratio is as
calculated below :-

                accp. ratio
2038 0.16304
2007 0.16056
2065 0.1652
2117 0.16936
2182 0.17456
2587 0.20696
3022 0.24176
3213 0.25704
3554 0.28432
3703 0.29624
0.211904 (avg accp ratio) (Is this value correct ??)

The Pdes used while generating temp. range was also 0.2. Does that mean
that  replicas have exchanged for the given temp.range ??.  Here's the link
for both remd_temp and remd_index files (
https://www.dropbox.com/s/uvwsdqjix49lg93/remd_index.png<https://www.dropbox.com/s/uvwsdqjix49lg93/remd_index.png;cid=1368069857486-810>)
, (
https://www.dropbox.com/s/78vcnaxzpgeekti/remd_temp.png?m<https://www.dropbox.com/s/78vcnaxzpgeekti/remd_temp.png?m&cid=1368069857486-810>
)

-- 
Bharat