Re: PMF Calculation

From: Brian Radak (bradak_at_anl.gov)
Date: Sun Aug 14 2016 - 16:36:06 CDT

I've never actually computed a PMF in this way, only free energies, but
this doesn't seem particularly surprising.

Ten realizations is not very many and fluctuation quantities (like the
second cumulant) are very hard to converge, even with a lot of data.
Collecting more data will likely only improve your results. You could
also analyze the estimated error by using a bootstrapping procedure, for
example.

HTH,

Brian

On 08/12/2016 07:15 PM, Crystal Liu wrote:
> Hello!
> I am using the same method from the deca-alanine tutorial
> (http://www.ks.uiuc.edu/Training/Tutorials/science/10Ala-tutorial/10Ala-tutorial.pdf
> <http://www.ks.uiuc.edu/Training/Tutorials/science/10Ala-tutorial/10Ala-tutorial.pdf>)
> to calculate PMF using Jarzynski's method using ten repeats of work
> data in kcal/mol; however, only the first cumulant (F1) plot worked,
> while the second cumulant (F2) plot gave me very irregular results,
> when both F1 and F2 are supposed to be similar. Can anyone point out
> where the problem might be? Thank you so much!
>
> /Equation:/
> Inline image 3
>
> B= 1/kT (k=Boltzmann's constant, T=temperature - at 310K)
>
>
> /Cumulant Calculation Scg ript:/
>
> set F1 {}
>
> set F2 {}
>
> for {set i 0} {$i < 75001} {incr i 1} {
>
> set texp 0
>
> set t1 0
>
> set t2 0
>
> foreach l [array names w] {
>
> set e [lindex $w($l) $i]
>
> set t1 [expr $t1 + $e]
>
> set t2 [expr $t2 + $e * $e]
>
> }
>
> set B [expr 1 / 1.2]
>
> lappend F1 [expr $t1 / 10]
>
> lappend F2 [expr $t1 / 10 - ($B * $t2 / 10) + ($B * $t1 * $t1 /
> 100) ]
>
> }
>
>
>
> /First cumulant in green, and second cumulant in blue: /
>
> Inline image 5
>

-- 
Brian Radak
Postdoctoral Appointee
Leadership Computing Facility
Argonne National Laboratory
9700 South Cass Avenue, Bldg. 240
Argonne, IL 60439-4854
(630) 252-8643
brian.radak_at_anl.gov

This archive was generated by hypermail 2.1.6 : Tue Dec 27 2016 - 23:22:22 CST