[gmx-users] Tabulated potential, derivative errors and interpolation

[ms]

David van der Spoel ha scritto:
> On 1/7/10 9:00 PM, ms wrote:
>> Update: It seems that, fixing the attractive values, the repulsive
>> values are well-behaved as well and the large errors disappear.
>>
>> Sorry for bothering you! I was fixated on the repulsion and I didn't
>> notice the blunder on the attraction...
>>
>> However it would be nice to know, just for the record, why errors in the
>> attraction lead to the weird oscillations in the repulsion...
>>
>> cheers,
>> M.
> Can you please file a bugzilla? This behavior does not seem correct.
> Please upload the necessary files to reproduce the problem.
> 

By the way, there's one thing bewildering me now.
*All* the signs I had before were wrong, possibly, because if I remember
correctly, in the .xvg table one has to put *minus* the derivative, and
for a repulsion term, this is expected to be a *positive* value (the
function will be monotonous decrescent).

Yet rtab14.xvg gives me *negative* values there.

Is the convention of rtab and of the tables different?

thanks,
m.