how to set internal parameter in specx.f for different system

Posted on : May 06, 2015 (Wed) 20:57:12

by Bin

Dear all
In case of the settings of internal parameter in specx,I found some reference in old version of AkaiKRR( in specx.f of cpa2002v009,June 27,2010) ,it is clear to set "natmmx", "ncmpmx" "miszmx", "mxlmx" according to the system size that we are going to compute,but how to set the parameter "nk1x" "nk3x" properly is not well told.( I think it is important to the final result of total energy since "nk1x" set a limit for maximum number of K-points for BZ intergration.)
Could anyone explain to me? thank you!

Below are the setting reference i mentioned above.
sample for natm=2, ncmpmx and msizmx are not uniqe.
c & (natmmx=14, ncmpmx=10, msizmx=140, mxlmx=4, nk1x=30, nk3x=21,
c & msex=201, ngmx=15, nrpmx=250, ngpmx=250, npmx=200, msr=400)
c--- sample for natm=6, ncmpmx and msizmx are not unique.
c & (natmmx=6, ncmpmx=4, msizmx=54, mxlmx=3, nk1x=130, nk3x=21,
c & msex=201, ngmx=15, nrpmx=250, ngpmx=250, npmx=200, msr=400)
c--- sample for natm=8, ncmpmx and msizmx are not unique.
c & (natmmx=16, ncmpmx=5, msizmx=96, mxlmx=3, nk1x=250, nk3x=21,
c & msex=201, ngmx=15, nrpmx=1600, ngpmx=1600, npmx=200, msr=400)
c--- sample for natm=10, ncmpmx and msizmx are not unique.
c & (natmmx=4, ncmpmx=9, msizmx=36, mxlmx=4, nk1x=1800, nk3x=21,
c & msex=201, ngmx=15, nrpmx=250, ngpmx=250, npmx=200, msr=400)
c--- sample for natm=4, ncmpmx and msizmx are not unique.
c & (natmmx=4, ncmpmx=9, msizmx=36, mxlmx=3, nk1x=550, nk3x=21,
c & msex=201, ngmx=15, nrpmx=250, ngpmx=250, npmx=200, msr=400)
c--- sample for natm=16, ncmpmx and msizmx are not unique.
c & (natmmx=27, ncmpmx=20, msizmx=211, mxlmx=3, nk1x=75, nk3x=1,
c & msex=151, ngmx=15, nrpmx=600, ngpmx=600, npmx=200, msr=400)
c--- sample for natm=36, ndmx is set by hand for (Fe,Co)(S,Se)2.
c & (natmmx=20, ncmpmx=7, msizmx=180, mxlmx=3, nk1x=24, nk3x=1,
c & msex=201, ngmx=15, nrpmx=250, ngpmx=250, npmx=200, msr=400)

[Re:01] how to set internal parameter in specx.f for different system

Posted on : May 06, 2015 (Wed) 23:54:04

by Hitoshi GOMI

Hello Bin,

The number of k-point (nk) is depending on the bzqlty in the input file (and also depending on the crystal structure).
nk is written in the output file. For example, in your previous calculation ( ttp://kkr.phys.sci.osaka-u.ac.jp/bbs/thread.cgi?id=6607 ), bzqlty is set to be 4, and resulting nk is 35.

nk1x+nk3x must be higher than nk. Otherwise, the execute file says "too many k-point".

Hitoshi

[Re:02] how to set internal parameter in specx.f for different system

Posted on : May 07, 2015 (Thu) 17:12:02

by Bin

Hitoshi,thank you for your reply,
I got your idea,You are always so helpful.