Re: ABF fullsamples

From: Jerome Henin (
Date: Mon Dec 03 2007 - 18:13:40 CST

Hi Jason,

On Dec 3, 2007 2:18 PM, Jason O'Young <> wrote:
> Hi All,
> I have another ABF question.
> How does the fullsamples parameter work? Is the ABF applied after ALL
> bins have accumulated the number of samples specified? Or is it only
> required of the current bin being sampled? I tired looking at the
> source code and I couldn't figure it out...

The number used is just the number of samples in the current bin: the
goal is to have a reasonable *local* estimate of the bias. The
situation is a little more complicated when smoothing is used, in
which case an effective number of samples involving neighboring bins
is used.
Also, there is a linear ramp: a scaled-down force starts being applied
when the number of samples reaches half of fullSamples, and the
scaling factor reaches 1 when Nsamples = fullSamples. This avoid
sudden jumps in the biasing force. The documentation should definitely
be more explicit on this...

> Additionally, is there a way to output the Fabf applied?

Not currently, but it would be easy to add. I don't have time to do it
now, but you could add a print statement for variable $F at line 294
in abf_script.tcl, that should do it. You may want to set an output
frequency to avoid flooding the output: just test the variable


This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:39 CST