[gmx-users] REMD average acceptance ratio

XAvier Periole x.periole at rug.nl
Sat May 11 20:36:07 CEST 2013 

The values of exchange ratio look much better: they are similar over the range of temperatures. 

To reduce the ratio you have to increase the spacing between temperatures, which means increase the value of c in the formula I gave earlier. 

When you but the water in, the degrees of freedom (and thus the potential energy) will increase and the temperature distribution won't work at all! To cover the same temperature range you'll need much more temperatures. 

On May 11, 2013, at 19:59, bharat gupta <bharat.85.monu at gmail.com> wrote:

> Dear Sir,
> 
> I tried again with the following temp. ditribution, this time with 30
> replicas (280 K -624K) and 500 ps simulation time for each one.
> 
> 0 280
> 1 287.8
> 2 295.9
> 3 304.2
> 4 312.7
> 5 321.5
> 6 330.5
> 7 339.8
> 8 349.3
> 9 359.1
> 10 369.1
> 11 379.5
> 12 390.1
> 13 401.0
> 14 412.3
> 15 423.8
> 16 435.7
> 17 447.9
> 18 460.4
> 19 473.3
> 20 486.6
> 21 500.2
> 22 514.3
> 23 528.7
> 24 543.5
> 25 558.7
> 26 574.4
> 27 590.4
> 28 607.0
> 29 624.0
> 
> 
> The output of   md29.log file is :-
> 
> Replica exchange statistics
> Repl  249 attempts, 125 odd, 124 even
> Repl  average probabilities:
> Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
>  14   15   16   17   18   19   20   21   22   23   24   25   26   27   28
>  29
> Repl      .68  .63  .62  .68  .71  .66  .69  .68  .71  .67  .64  .71  .69
> .71  .66  .69  .73  .73  .72  .73  .69  .71  .71  .74  .73  .70  .72  .74
> .71
> Repl  number of exchanges:
> Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
>  14   15   16   17   18   19   20   21   22   23   24   25   26   27   28
>  29
> Repl       81   77   79   90   93   78   91   80   88   78   81   92   96
> 94   81   83   90   92   83   97   89   87   91   94   88   84   85   89
> 86
> Repl  average number of exchanges:
> Repl     0    1    2    3    4    5    6    7    8    9   10   11   12   13
>  14   15   16   17   18   19   20   21   22   23   24   25   26   27   28
>  29
> Repl      .65  .62  .63  .73  .74  .63  .73  .65  .70  .63  .65  .74  .77
> .76  .65  .67  .72  .74  .66  .78  .71  .70  .73  .76  .70  .68  .68  .72
> .69
> 
> 
> The average acceptance ration comes around 0.69 , which is very high. Now,
> in order to get the avg. acceptance ration between 0.2 to 0.3 and also all
> the replicas should exchange , what has to be done.
> 
> Here's the link for remd_index and remd_temp files
> https://www.dropbox.com/s/sgpblcdg9zh7f52/remd_temp.png
> https://www.dropbox.com/s/6hvqlqmu64mb2jy/remd_index.png
> 
> I want to that if I include water for the simulation, the same temp.
> distribution would work or not ??
> 
> 
> 
> 
> On Sun, May 12, 2013 at 12:10 AM, XAvier Periole <x.periole at rug.nl> wrote:
> 
>> 
>> You are simulating in vacuo! Otherwise the temperature gaps are way too
>> large …
>> 
>> If you want to analyse the sampling at 300 K, I would suggest you start
>> you first temperature lower, around 280/285 may be. At least to have your
>> second temperature at 300 K.
>> 
>> the value of c has absolutely not importance … the temperature
>> distribution has … make some test to see how the acceptance ratio evolves …
>> 
>> On May 11, 2013, at 5:05 PM, bharat gupta <bharat.85.monu at gmail.com>
>> wrote:
>> 
>>> Dear Sir,
>>> 
>>> Here's the temperature range that I got form t-remd :
>>> 1 300
>>> 2 323.7
>>> 3 348.75
>>> 4 375.23
>>> 5 403.22
>>> 6 432.83
>>> 7 464.14
>>> 8 497.24
>>> 9 532.26
>>> 10 569.32
>>> 11 608.51
>>> 
>>> 
>>> according the above equation c should be somewhere around 2.37.
>>> 
>>> 
>>> On Sat, May 11, 2013 at 11:10 PM, XAvier Periole <x.periole at rug.nl>
>> wrote:
>>> 
>>>> 
>>>> Well, actually things do not look so good. But is it possible that grace
>>>> is actually no able to plot things correctly? You have line going
>>>> throughout the plot from complete-left to complete-right!
>>>> 
>>>> I am do not know what the t-rems calculator does but apparently it is
>> not
>>>> optimal in your case. Did you try to use the simple rule Tn=T0 x exp(n
>> c),
>>>> where T0 is close to your starting temperature and c is a constant that
>> you
>>>> can tune and will define the spacing between the temperatures. From your
>>>> current data you can guess the spacing and thus the c value you need.
>> Note
>>>> that the exchange ratio is quickly converging in the simulation so you
>> can
>>>> make a few trials …
>>>> 
>>>> On May 11, 2013, at 1:40 PM, bharat gupta <bharat.85.monu at gmail.com>
>>>> wrote:
>>>> 
>>>>> Dear Sir,
>>>>> 
>>>>> Thank you for your reply. I choose the temperature distribution using
>>>>> t-remd calculator. Here's the link for index and temp files .
>>>>> https://www.dropbox.com/s/uvwsdqjix49lg93/remd_index.png
>>>>> https://www.dropbox.com/s/78vcnaxzpgeekti/remd_temp.png?m.
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> 
>>>>> On Sat, May 11, 2013 at 12:04 AM, XAvier Periole <x.periole at rug.nl>
>>>> wrote:
>>>>> 
>>>>>> 
>>>>>> The replicas seem indeed to have exchange. Using a colour for the #
>>>>>> replicas would help.
>>>>>> 
>>>>>> I could not access to the first link.
>>>>>> 
>>>>>> Note also that the increase of exchange ratio with the temperature
>>>> suggest
>>>>>> the distribution of the temperature is not optimal and may be with
>>>> regular
>>>>>> intervals? You want to use a exponential distribution.
>>>>>> 
>>>>>> On May 10, 2013, at 4:53 PM, bharat gupta <bharat.85.monu at gmail.com>
>>>>>> wrote:
>>>>>> 
>>>>>>> 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
>>>>>>> --
>>>>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>>>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>>>>>> * Please search the archive at
>>>>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>>>>>> * Please don't post (un)subscribe requests to the list. Use the
>>>>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>>> 
>>>>>> --
>>>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>>>>> * Please search the archive at
>>>>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>>>>> * Please don't post (un)subscribe requests to the list. Use the
>>>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>>> 
>>>>> 
>>>>> 
>>>>> --
>>>>> Bharat
>>>>> Ph.D. Candidate
>>>>> Biomolecular Engineering Laboratory
>>>>> Pusan National University
>>>>> South Korea
>>>>> Mobile no. - 010-5818-3680
>>>>> --
>>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>>>> * Please search the archive at
>>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>>>> * Please don't post (un)subscribe requests to the list. Use the
>>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>> 
>>>> --
>>>> gmx-users mailing list    gmx-users at gromacs.org
>>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>>> * Please search the archive at
>>>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>>> * Please don't post (un)subscribe requests to the list. Use the
>>>> www interface or send it to gmx-users-request at gromacs.org.
>>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>> 
>>> 
>>> 
>>> --
>>> --
>>> gmx-users mailing list    gmx-users at gromacs.org
>>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>>> * Please don't post (un)subscribe requests to the list. Use the
>>> www interface or send it to gmx-users-request at gromacs.org.
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> 
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/mailman/listinfo/gmx-users
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
>> * Please don't post (un)subscribe requests to the list. Use the
>> www interface or send it to gmx-users-request at gromacs.org.
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> -- 
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.gromacs.org/mailman/listinfo/gmx-users
> * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting!
> * Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

More information about the gromacs.org_gmx-users mailing list