comp-chem/NWChem
1
0
Fork 0
mirror of https://github.com/nwchemgit/nwchem.git synced 2026-07-28 14:15:30 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

include info about possible shrinking of radial grids when using

Browse source 
lebedev angular grids.
This commit is contained in:
Jeff Nichols 1999-07-30 00:43:51 +00:00
parent 03a5dd85f3
commit a3b85ba2a3
1 changed files with 17 additions and 16 deletions
Download patch file Download diff file Expand all files Collapse all files

33
doc/user/dft.tex
View file

@ -694,22 +694,23 @@ to generate the desired energy accuracy (with utter disregard for speed).
{\bf Important note to users.} We clearly understand that the default
(Euler-MacLaurin/Gauss-Legendre) grids are large and result in slow
construction of the numerical components of the Kohn-Sham equations.
Alternatively, we have provided access to two-dimensional Lebedev angular
quadratures which can be used in many
cases to substantially reduce the number of grid points per atom while
keeping good accuracy. We have not yet had the opportunity to benchmark
the Lebedev angular quadratures to the same extent that we have for
the Gauss-Legendre. We therefore do not have recommended Lebedev
quadratures for specific target accuracy for all elements of the
periodic table. If the user wants to significantly decrease CPU time
to solution it is suggested that a few prototype benchmark
calculations be done using various Lebedev quadratures (which we describe
below) while
monitoring the numerically integrated density and total energies for
the molecular systems of interest. For many examples we have observed
speed-ups of two or more for the same numerical
accuracy when using Lebedev rather than the default Gauss-Legendre
quadrature.
Alternatively, we have provided access to two-dimensional Lebedev
angular quadratures which can be used in many cases to substantially
reduce the number of grid points per atom while keeping good accuracy.
We have not yet had the opportunity to benchmark the Lebedev angular
quadratures to the same extent that we have for the Gauss-Legendre.
We therefore do not have default Lebedev quadratures for specific
target accuracy for all elements of the periodic table. If the user
wants to significantly decrease CPU time to solution it is suggested
that a few prototype benchmark calculations be done using various
Lebedev quadratures (which we describe below) while monitoring the
numerically integrated density and total energies for the molecular
systems of interest. For many examples we have observed speed-ups of
two or more for the same numerical accuracy when using Lebedev rather
than the default Gauss-Legendre quadrature. In addition, we have
observed that with Lebedev angular quadratures a reduction in the
number of radial shells (perhaps by as much as 30\%) might be possible
while continuing to provide the same level of accuracy.
\begin{table}[h]