Hi,
I have a problem with the error analysis of g_wham. The PMF profile and
sampling windows looks good, but when I try to generate error bars g_wham
hangs. This is what I do:
g_wham_mpi -if pullf-files.dat -it tpr-files.dat -o -hist -nBootstrap 50 -bins
50 -bs-method traj-gauss -bsres -ac y
On 8/5/13 2:29 AM, Magnus Andersson wrote:
Hi,
I have a problem with the error analysis of g_wham. The PMF profile and
sampling windows looks good, but when I try to generate error bars g_wham
hangs. This is what I do:
g_wham_mpi -if pullf-files.dat -it tpr-files.dat -o -hist -nBootstrap 5
>
> I've had this happen before when trying to analyze too many data points. Try
> reducing the size of the input data using -b to see if that helps.
>
> -Justin
Thanks for the suggestion, Justin. However, either it hangs or I get:
Fatal error:
Tryig to estimtate autocorrelation time from on
On 8/5/13 8:06 AM, Magnus Andersson wrote:
I've had this happen before when trying to analyze too many data points. Try
reducing the size of the input data using -b to see if that helps.
-Justin
Thanks for the suggestion, Justin. However, either it hangs or I get:
Fatal error:
Tryig to
> What version of Gromacs are you using? How many data points do you have in
> the full set?
>
> -Justin
>
Gromacs 4.6
not sure what you mean by data points, but:
dt = 0.002
nsteps = 100
and total number of atoms in the system: 265,891
# frames
On 8/5/13 8:21 AM, Magnus Andersson wrote:
What version of Gromacs are you using? How many data points do you have in the
full set?
-Justin
Gromacs 4.6
not sure what you mean by data points, but:
dt = 0.002
nsteps = 100
Not relevant, but
Hi Magnus,
I just read your mail. Sound very much like a bug, which never occurred
to me though. But I did not use -bs-method traj-gauss a lot. Does the
error also occur with -bs-method traj?
Btw: with -bs-method traj-gauss or traj, you need realistic estimates of
the autocorrelation times (ACTs)
Hi Jochen,
Seems as if this error was associated with the mpi version. Everything worked
ok when running it locally.
Best regards /
Magnus
On Aug 12, 2013, at 1:50 PM, Jochen Hub wrote:
> Hi Magnus,
>
> I just read your mail. Sound very much like a bug, which never occurred
> to me though.
8 matches
Mail list logo