Re: [SIESTA-L] WARNING: High Pressure calculation!
Dear Siesta users, recently Andrei Postnikov wrote a small tutorial on Siesta http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf He wrote: one cannot a priori expect equally good performance of basis over a large range of bond lengths and, say, pressures: as the pressure is varying, the “completeness” or “variational freedom” of basis in the region of chemically relevant, under a given pressure, bond lengths do vary; consequently the calculated pressure/energy curve may slightly differ from that obtained in a “benchmark” FLAPW calculation. A good test would be to compare bulk VASP and Siesta calculations and deside if calculation in Siesta is accurate enough at some particular pressure. Unfortunately we don't have VASP here. My question is (+/- 10%) of the equlibrium lattice constants of Si and Ge are Ok for Siesta? Will be Siesta accurate enough to calculate surface energy with an error of about 1 meV/A2 ? Dear Ruslan, this seems to be a too loosely defined question; will be Siesta accurate enough with which basis, cutoffs etc.? It would demand a more careful study, volume curve vs. volume curve, in the spirit or Junquera's tests of basis - Fig.2 of PRB64, 235111. If you don't have VASP nor WIEN2k, install ABINIT or PWSCF which should be good enough for Si and Ge. The surface energy is another philosophical question (the issue of wrong asymptotics on the surface? Using different pseudo?) - I think this has been addressed in some of quite early Siesta papers. Best regards Andrei Postnikov
Re: [SIESTA-L] WARNING: High Pressure calculation!
Dear Andrei, this seems to be a too loosely defined question; will be Siesta accurate enough with which basis, cutoffs etc.? basis set DZP mesh cutoff 200 Ry I did test calculations for Si(100) (unreconstructed and unrelaxed) surface energy in Siesta, the Esurf is very close to published values obtained in VASP (error less than 1 meV/A2). So I am quite sure about this. Kind regards Ruslan 2009/6/4 apost...@uni-osnabrueck.de Dear Siesta users, recently Andrei Postnikov wrote a small tutorial on Siesta http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf He wrote: one cannot a priori expect equally good performance of basis over a large range of bond lengths and, say, pressures: as the pressure is varying, the “completeness” or “variational freedom” of basis in the region of chemically relevant, under a given pressure, bond lengths do vary; consequently the calculated pressure/energy curve may slightly differ from that obtained in a “benchmark” FLAPW calculation. A good test would be to compare bulk VASP and Siesta calculations and deside if calculation in Siesta is accurate enough at some particular pressure. Unfortunately we don't have VASP here. My question is (+/- 10%) of the equlibrium lattice constants of Si and Ge are Ok for Siesta? Will be Siesta accurate enough to calculate surface energy with an error of about 1 meV/A2 ? Dear Ruslan, this seems to be a too loosely defined question; will be Siesta accurate enough with which basis, cutoffs etc.? It would demand a more careful study, volume curve vs. volume curve, in the spirit or Junquera's tests of basis - Fig.2 of PRB64, 235111. If you don't have VASP nor WIEN2k, install ABINIT or PWSCF which should be good enough for Si and Ge. The surface energy is another philosophical question (the issue of wrong asymptotics on the surface? Using different pseudo?) - I think this has been addressed in some of quite early Siesta papers. Best regards Andrei Postnikov
Re: [SIESTA-L] WARNING: High Pressure calculation!
This basis set completeness problem is the basis set superposition error (BSSE), which has been thoroughly studied in molecular calculations. All methods based on atom centred basis sets suffer from it. It basically leads to an artificial attractive force between atoms, because in a more compact system, each atom is described with a more complete basis. In the context of periodic calculations, I would particularly expect it to overestimate surface adsorption energies and underestimate bond lengths. Cheers, Herbert ?? ? wrote: Dear Siesta users, recently Andrei Postnikov wrote a small tutorial on Siesta http://www.home.uni-osnabrueck.de/apostnik/Lectures/SIESTA-tuto.pdf He wrote: one cannot a priori expect equally good performance of basis over a large range of bond lengths and, say, pressures: as the pressure is varying, the “completeness” or “variational freedom” of basis in the region of chemically relevant, under a given pressure, bond lengths do vary; consequently the calculated pressure/energy curve may slightly differ from that obtained in a “benchmark” FLAPW calculation. A good test would be to compare bulk VASP and Siesta calculations and deside if calculation in Siesta is accurate enough at some particular pressure. Unfortunately we don't have VASP here. My question is (+/- 10%) of the equlibrium lattice constants of Si and Ge are Ok for Siesta? Will be Siesta accurate enough to calculate surface energy with an error of about 1 meV/A2 ? Kind regards Ruslan Zhachuk -- Herbert Fruchtl Senior Scientific Computing Officer School of Chemistry, School of Mathematics and Statistics University of St Andrews -- The University of St Andrews is a charity registered in Scotland: No SC013532
