[gmx-users] Coulomb barriers and Coulomb Softcore Potential

Wed Aug 26 14:44:00 CEST 2015

Dear Mr. Shirts,

thanks for your reply. My N/N_k ratio looks like this:

Number of correlated and uncorrelated samples:

  State            N          N_k        N/N_k
      0       600001        31483        19.06
      1       600001        31122        19.28
      2       600001        34961        17.16
      3       600001         9374        64.01
      4       600001         2249       266.86
      5       600001          532      1129.04
      6       600001           94      6384.18
      7       600001          485      1237.84
      8       600001          585      1027.03
      9       600001         1684       356.43
     10       600001         1047       573.48
     11       600001         8207        73.11
     12       600001          347      1731.24
     13       600001         2702       222.13
     14       600001           79      7692.14
     15       600001          193      3120.28
     16       600001          414      1451.40
     17       600001        57985        10.35
     18       600001        60589         9.90
     19       600001        64621         9.28

With the following lambda distribution:

; init_lambda_state        0     1     2     3     4     5     6 7     
8     9     10    11    12    13    14    15    16    17 18 19
vdw_lambdas             =  0.000 0.224 0.378 0.489 0.578 0.657 0.730 
0.802 0.881 0.976 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 
1.000
coul_lambdas            =  0.000 0.000 0.000 0.000 0.000 0.000 0.000 
0.000 0.000 0.000 0.100 0.210 0.303 0.383 0.452 0.524 0.604 0.712 0.853 
1.000

and lambda values 0 = decoupled / 1 = coupled.

I use the LJ 1-1-6 SCP with alpha = 0.7 (I've checked alpha = 0.5 as 
well, but there is no significant difference), while Coulomb Softcore is 
not active.

I've uploaded a diagram of Gibbs free energy difference over simulation 
time on github (https://github.com/AMecklenfeld/Ionic-Liquid/issues/1). 
The diagram contains some plateaus with high steps and slowly 
approaches  a final value. From my point of view, it actually looks like 
my system is this slow.

My idea was to use the Coulomb-SCP to flatten the energetic landscape. 
Another way would be expanded ensemble simulation, but since I use a 
constrained 4-site water model (triangle constraints), I've difficulties 
with shake in Gromacs 5.0.4.

Do you have any suggestions?

With kind regards,
Andreas Mecklenfeld

Am 24.08.2015 um 17:08 schrieb Michael Shirts:
> Adjusting the soft core is a gigantic pain.  I wouldn't recommend it,
> and it's likely not necesssary. What settings are you using now?  Note
> that the manual describes how it is defined.
>
> Can you post the alchemical-analysis output for N/N_k?  There's an
> analysis quirk we are looking at where the correlation times are
> measured to be longer than they actually are, though that is when LJ
> is changing and coul is turned off.
>
> Look at the time correlation for the variable that is changing -- does
> it look stationary, or does it look like it's changing slowly, or has
> just a few discrete steps?  If the autocorrelation time is slow, then
> N/N_k actually is that long, and the code is doing it's job.
>
> On Mon, Aug 24, 2015 at 9:55 AM, Andreas Mecklenfeld
> <a.mecklenfeld at tu-braunschweig.de> wrote:
>> Dear Gromacs-Users,
>>
>> I want to calculate the solvation free energy of water in an aqueous ionic
>> solution and I'm using the Python tool "alchemical-analysis.py" by
>> Klimovich, Shirts and Mobley for evaluation. This tool demonstrates a very
>> high N/N_k ratio (up to 7000) while decreasing the electrostatic potential
>> (Lennard Jones fully active).
>>
>> Considering energetic barriers, I would like to adjust the Coulomb Softcore
>> Potential. Does this seem plausible and if so, how is the Coulomb Softcore
>> Potential defined in Gromacs?
>> Naden and Shirts provide a concept by equation (A.2) in "Linear Basis
>> Function Approach to Efficient Alchemical Free Energy Calculations. 2.
>> Inserting and Deleting Particles with Coulombic Interactions"
>> (http://pubs.acs.org/doi/abs/10.1021/ct501047e) - is this the formula
>> intended?
>>
>> Best regards,
>> Andreas
>>
>> --
>> M. Sc. Andreas Mecklenfeld
>> Stipendiat
>>
>> Technische Universität Braunschweig
>> Institut für Thermodynamik
>> Hans-Sommer-Straße 5
>> 38106 Braunschweig
>> Deutschland / Germany
>>
>> Tel: +49 (0)531 391-2634
>> Fax: +49 (0)531 391-7814
>>
>> www.ift-bs.de
>>
>> --
>> Gromacs Users mailing list
>>
>> * Please search the archive at
>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
>>
>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>
>> * For (un)subscribe requests visit
>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
>> mail to gmx-users-request at gromacs.org.

-- 
M. Sc. Andreas Mecklenfeld
Stipendiat

Technische Universität Braunschweig
Institut für Thermodynamik
Hans-Sommer-Straße 5
38106 Braunschweig
Deutschland / Germany

Tel: +49 (0)531 391-2634
Fax: +49 (0)531 391-7814

http://www.ift-bs.de