[Pw_forum] PWscf and highly parallel machines

[Paolo Giannozzi]

On Feb 15, 2007, at 16:37 , Nichols A. Romero wrote:

> There are two ways to solve for the KS ground state.
> 1. Direct minimization of KS functional
> 2. Self-consistently solving the KS eqn.
>
> The conjugate-gradient (cg) in PWSCF and CP are not doing
> different things same thing.

let me say something different the same thing :-)

'cg' in PWscf: *self-consistency + fixed-potential diagonalization*
The diagonalization is performed by transforming the secular
problem into a minimum problem (nothing really fancy:
min <\psi|H\psi> + orthogonality) and solving for each band
in sequence, using conjugate gradient. Slower than Davidson
diagonalization, uses less memory. To be used in case of trouble
with Davidson or if memory is tight.

'cg' in CP: conjugate-gradient algorithm for minimization of the
*energy functional* (global minimization).

Paolo
---
Paolo Giannozzi, Democritos and University of Udine, Italy