Re^2:how to run a system of over 30 atoms?

Posted on : April 14, 2007 (Sat) 10:27:57

by zhoul

Thank you for your patient guide.

Now I have successfully set the parameters as the following cases:
natmmx=32, ncmpmx=5, msizmx=300, mxlmx=3,nk1x=20,nk3x=21

And in the test input file ,I have 32 atoms and "bzqlty=1".

Howerer ,it couldn't be done successfully.The error message says
"***err in madlng ... negtive volint"

Could it be something wrong with my input file?

Besides,it still could be done when I decresed the number of atoms to about 6.


>It is still possible to run for a system with over 30 atoms.
>Take care about the following:
>
>1) Decrease the number of k-points in order that the memory
>addressing does not overflow (in the case of 32bit addressing
>system, such as Pentium 4 with 32bit addressing). This can be
>done by decreasing "nk1x". The test run should be done
>bu "bzqlty=0" or "bzqlty=1" for which nk1x=1~20 would
>be enough.
>
>2) Increase the number aoms that can be handled:
>For that increase "natmmx", "ncmpmx", and "msizmx".
>
>For example if you have 30 atoms and for each atom
>you would like to calculate up to d states (l=2),
>all atoms have different types, each site has a single component
>(i.e. not a random alloy) the following parameters are suitable:
>natmmx=30, ncmpmx=30, msizmx=270, mxlmx=3