The ABINIT Users Mailing List ( CLOSED )

[<g930102@ccit.edu.tw>]

Some comment about GW calculation in ABINIT

Dear Dr. Gian-Marco:

I am study the topics of the GW calculation in solids at present.
Except the ABINIT( GW calculation based on PP & PW, Plane wave
and PseudoPotentials),I am also use the DFTGWA (GW calculation 
based on All Electrons FP-LMTO). I have ported DFTGWA(fpgw020)
into Intel(Compiler)/Linux(OS)/PC(Vendor) version in my own site.
In my points of view, DFTGWA is GW calculation exclusively. 
You could to do execution LMTO and GW separately. The scheme some
like as GWA in ABINIT before V3.4. So I have no difficulty to 
switch each one, even though there are drastically different from
input scheme(Lattice coordinates and Cartesian coordinates, 
for example). 

I think not all of peoples need to GW calculation in his work. 
Because of "computing GW corrections is a quite heavy calculations"
(cited from Mike Zhang and Gian-Marco Rignanese communications). 
In my experiments, it is count by day for GW calculations for both 
packages(DFTGWA and ABINIT).
(ZincBlende: MP mesh 8x8x8, 29 k-points, 512 k-points in IBZ,
29 q-points. WurtZite: MP mesh 6x6x4, 21 k-points, 144 k-points in 
IBZ, 21 q-points. May could covers full band calculations in 
coarsely?) (DFTGWA use the tetrahedron approach which is conventional
method for LMTO)

I am expect the ABINIT V3.4x could work in P6/DOS/Windows. I have 
triedto work it out, but I am failed in memory allocation. This is 
caused by GW code integration with ABINIT, needs more arrays space. 
That is what I am encountered. So I wonder "Availability's Policy 
Change: Do not support sequential version in Dos/Win after 3.4?", 
I think Dr. Luc Sindic have much heavy work in the near future.

I am sorry that I have not test run V3.4x in Linux? May be it will 
give a surprise to me. But to my knowledge, there will exhaust most 
resourcesin computer while GW code integration with ABINIT. May be 
I am too worry? 

Best regards,

Lu Fu-Fa   Sep/05/2002
=======================================================
Lu Fu-Fa, Doctoral Student, E-mail: g930102@ccit.edu.tw
Semiconductors Lab., Dept. of Applied Physics
Chung Cheng Institute of Technology (CCIT), TW
=======================================================

___________________________________
NOCC, http://nocc.sourceforge.net