In fact by default SFALL uses a grid of 0.5A or less even for low resln
data - this is because the atomic spread id limited to 2.%A by default,
so an atom can sort of fall through the grid if it is too coarse.
Eleanor
FFTs are so fast that the time for most cells is pretty insignificant..
Bernhard Rupp wrote:
what is the limit of N in a practical case for the Gaussian
approximation.........
The resolution. Less resolution, less terms needed.
see Acta papers TenEyck 1977 particularly Agarwal 1978,
and the cctbx section in newsletter:
http:// <http://cci.lbl.gov/publications/download/iucrcompcomm_jan2004.pdf>
cci.lbl.gov/publications/download/iucrcompcomm_jan2004.pdf
br
_____
From: Jayashankar [mailto:[EMAIL PROTECTED]
Sent: Thursday, October 11, 2007 10:54 AM
To: [EMAIL PROTECTED]
Subject: Re: [ccp4bb] atomic FF used in SFALL
On 10/11/07, Bernhard Rupp < <mailto:[EMAIL PROTECTED]>
[EMAIL PROTECTED]> wrote:
It seems that 5-Gaussian is indeed the accepted technical term
for the 9-parameter expansion:
Given that a5=c and b5=0, the 5th Gaussian can be a constant.
Continuing this reasoning, even a binary mumber can be represented
as series of Gaussians.....a=0,b=0; a=1, b=0
This is called 'Gomputing'.
Cheers, br
-----Original Message-----
From: Eleanor Dodson [mailto:[EMAIL PROTECTED]
Sent: Thursday, October 11, 2007 1:26 AM
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: Re: [ccp4bb] atomic FF used in SFALL
I wrote that - is a constant a Gaussian? Anyway that is how I got 5 - and it
is the
9-parameter Cromer-Mann approximation..
And Yes - each component is added to the B value to build up the real space
atomic density.
as you show.
Eleanor
