[CPMD-list] (no subject)

[Axel Kohlmeyer]

On Mon, 2 May 2005, Rad Balu wrote:

RB> Hi, 

hi rad,

RB> I am working on the second approach and when I get things working and
RB> tested against few systems I will provide the group with the scripts.
RB> I can't find enough details about the geo interface either in the
RB> documentation or in the code. There are couple of lines of comments in
RB> the code that is not even saying how to invoke the interface for
RB> properties such as dipoles (local).

you have to dig a little deeper to figure out the handshake.
basically it boils down (salomon please correct me, if i am
getting it wrong).

1.) if the name of the input file contains .run, then cpmd
waits on start until a file QMSTART appears. this is supposed
to be written by the 'driver' program

-> control.F around line 338

cpmd will delete the file and thus the driver knows, that
cpmd is running.

2.) the rest of the handshake should be in egointer.F
where the two codes tell each other to get started by
creating a file name QMCONTINUE or EGOCONTINUE respectively

RB> I am also running the MD runs (dipole dynamics) for a system with 56 atoms.
RB> So far it has accumulated around 50000 dipole data points. When I run the
RB> fourier program that I downloaded from the tutorial web site, I only get
RB> about 25 - 30 frequencies. Do I have do accumulate lot more points for such
RB> a calculation and if so, roughly how many more points may be required?

you probably have a continued run. those have lines with '<< NEW DATA >>' 
the code posted on the tutorial page does not know how to handle those.
try removing those lines (best on a copy, if the job is still running).

RB> I am also exploring the package pwscf for intensity calculations. From the
RB> document for the package it appears that it supports IR intensity
RB> calculations and it also has cpmd as part of the package. What is the
RB> relation between these two packages ?(that is pwscf and cpmd) Is it possible
RB> to use the code in pwscf for intensity calculations with cpmd modules?

the CP-MD in pwscf (or quantum-espresso to be more precise) are actually
(still) two codes that implement the same car-parrinello hamiltonian
as cpmd does, also several developers worked at some point in time
on either or both codes and there may even be some stuff in them that
dates back to the first car-parrinello code. from a practical point of
view the CPMD code has its strong points in running CP molecular dynamics
at the gamma point with excellent parallel scaling, whereas the pwscf code 
has much better support for k-points (including ultrasoft pseudopotentials)
and metals. many of its features seem to be more geared towards solid 
state physics. of course this is an oversimplification and my personal
impression but that should at least represent the general trend.

to move code from pwscf to cpmd you have to be aware of the fact
that pwscf is licensend under GPL and cpmd is not. also pwscf is
(mostly) fortran 90 whereas cpmd is (mostly) fortran 77.
as far as i understand it, you may interface cpmd to pwscf, at 
the very least when you do it only for youself, but you should 
consult alessandro curioni about the implications when you want
to contribute such code for inclusion into the cpmd distribution.

RB> Can pwscf be compiled with intel compiler for IA64 and MPI architecture?

yes, pwscf works with IA64 and MPI. the intel compilers and MKL (needed
for IA64) are a primary development platform (on ia32, though).
for more details, please see the www.pwscf.org and there also is 
a pwscf mailing list. 

regards,
	axel.

RB> 
RB> Thanks in advance for providing any pointers.
RB> Rad

--
=======================================================================
Dr. Axel Kohlmeyer   e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.