Hi Mike
Yes you are right. increasing the number of digits for the format
in the files that are used as communication between various programs
should solve the problem, with accuracy up to that last digit.
But, of course, there are many other approximations on the way
that influence the quality of the resultant charges.
Best,
Piotr
On Fri, 2012-02-03 at 11:55, Crowley, Michael wrote:
> Hi Francois,
> Thanks a lot for your interest, I see your points.
> I am fine with lower precision, but do not understand what antechamber is
> doing with the lower precision. The charges do not make sense compared to
> the higher precisions, they are not rounded off numbers, they are
> different and don’t sum to zero, not even close. From what I see, the
> first column should have -0.0939 and 0.0313. Even with the -0.0941, the H
> charges should be 0.0314 not 0.0317. Even with two digits of precision,
> the charges should be -.094 and .031 with a sum of 0.001, half of the
> antechamber sum/error.
> formats from sqm
> f14.3 f14.4 f14.5 f14.8
> -0.094100 -0.093900 -0.093890 -0.093887
> -0.094100 -0.093900 -0.093890 -0.093887
> 0.031700 0.031300 0.031300 0.031296
> 0.031700 0.031300 0.031300 0.031296
> 0.031700 0.031300 0.031300 0.031296
> 0.031700 0.031300 0.031300 0.031296
> 0.031700 0.031300 0.031300 0.031296
> 0.031700 0.031300 0.031300 0.031296
> total
> 0.002000 0.000000 0.000000 0.000001
>
>
> Does that make sense, or am I still wanting too much from these programs?
> At least for this case it looks to me that just changing the precision by
> one or two in the printout is a workaround if not a real solution.
>
> Best wishes,
> Michael
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Feb 03 2012 - 13:30:03 PST
