Siesta

Nesta seção:

Descrição
Versões Disponíveis
Carregando o Módulo
Submissão de Jobs Seriais
Submissão de Jobs MPI
Exemplo de Arquivo de Entrada (FDF)
Otimização de Geometria
Cálculo de Bandas
Density of States
Job Array para Varredura de Parâmetros
Análise de Resultados
Referências

Descrição 

De acordo com a página do Siesta, SIESTA é um método de implementação computacional para realizar cálculos eficientes de estrutura eletrônica e simulações de dinâmica molecular ab initio de moléculas e sólidos, utilizando bases localizadas estritamente.

Versões Disponíveis 

siesta/4.0.2
siesta/4.1-b3 (default)
siesta/4.1.5.mpi
siesta/4.1.5.serial
siesta/5.2.0.mpi
siesta/5.2.0.serial

Carregando o Módulo 

# Versão serial
module load siesta/5.2.0.serial

# Versão MPI
module load siesta/5.2.0.mpi

Submissão de Jobs Seriais 

submit_siesta_serial.sh

#!/bin/bash
#SBATCH -J siesta_serial
#SBATCH -N 1
#SBATCH -n 1
#SBATCH -t 24:00:00
#SBATCH --mem=8G

export INPUT="h2o.fdf"
export OUTPUT="h2o.out"

module load siesta/5.2.0.serial

job-nanny siesta < h2o.fdf > h2o.out

Submissão de Jobs MPI 

submit_siesta_mpi.sh

#!/bin/bash
#SBATCH -J siesta_mpi
#SBATCH -N 2
#SBATCH --ntasks-per-node=28
#SBATCH -t 48:00:00
#SBATCH --mem-per-cpu=2G

export INPUT="si.fdf"
export OUTPUT="si.out"

module load siesta/5.2.0.mpi

job-nanny mpirun -np $SLURM_NTASKS siesta < si.fdf > si.out

Exemplo de Arquivo de Entrada (FDF)

h2o.fdf

# Siesta input for H2O molecule

SystemName          Water molecule
SystemLabel         h2o

# Lattice parameters (large box for molecule)
LatticeConstant     20.0 Bohr
%block LatticeVectors
  1.0  0.0  0.0
  0.0  1.0  0.0
  0.0  0.0  1.0
%endblock LatticeVectors

# Atomic coordinates
AtomicCoordinatesFormat  Ang
%block AtomicCoordinatesAndAtomicSpecies
  0.000   0.000   0.000  1  # O
  0.757   0.586   0.000  2  # H
 -0.757   0.586   0.000  2  # H
%endblock AtomicCoordinatesAndAtomicSpecies

# Atomic species
NumberOfSpecies        2
%block ChemicalSpeciesLabel
  1  8  O
  2  1  H
%endblock ChemicalSpeciesLabel

# Basis set
PAO.BasisSize      DZP
PAO.EnergyShift    0.02 Ry

# Exchange-correlation
XC.functional      GGA
XC.authors         PBE

# Mesh
MeshCutoff         200 Ry

# SCF options
MaxSCFIterations    200
DM.MixingWeight     0.1
DM.Tolerance        1.d-6

# Output
WriteForces         T
WriteCoorStep       T
WriteMDhistory      F

# Geometry optimization
MD.TypeOfRun        CG
MD.NumCGsteps       100
MD.MaxForceTol      0.01 eV/Ang

Otimização de Geometria 

submit_siesta_opt.sh

#!/bin/bash
#SBATCH -J siesta_opt
#SBATCH -N 1
#SBATCH -n 1
#SBATCH -t 36:00:00
#SBATCH --mem=8G

export INPUT="optimize.fdf"
export OUTPUT="optimize.out"

module load siesta/5.2.0.serial

job-nanny siesta < optimize.fdf > optimize.out

optimize.fdf

# Siesta geometry optimization

SystemName          Si bulk optimization
SystemLabel         si_opt

# Lattice parameters for Si
LatticeConstant     10.20 Bohr
%block LatticeVectors
  0.0  0.5  0.5
  0.5  0.0  0.5
  0.5  0.5  0.0
%endblock LatticeVectors

# Atomic positions
AtomicCoordinatesFormat  Fractional
NumberOfAtoms        2
%block AtomicCoordinatesAndAtomicSpecies
  0.00   0.00   0.00  1
  0.25   0.25   0.25  1
%endblock AtomicCoordinatesAndAtomicSpecies

NumberOfSpecies      1
%block ChemicalSpeciesLabel
  1  14  Si
%endblock ChemicalSpeciesLabel

# Basis
PAO.BasisSize        DZP
PAO.EnergyShift      0.02 Ry

# k-points
%block kgrid_Monkhorst_Pack
  4 0 0 0.0
  0 4 0 0.0
  0 0 4 0.0
%endblock kgrid_Monkhorst_Pack

# Optimization
MD.TypeOfRun         CG
MD.NumCGsteps        100
MD.MaxForceTol       0.01 eV/Ang

Cálculo de Bandas 

submit_siesta_bands.sh

#!/bin/bash
#SBATCH -J siesta_bands
#SBATCH -N 2
#SBATCH --ntasks-per-node=28
#SBATCH -t 48:00:00
#SBATCH --mem-per-cpu=2G

export INPUT="bands.fdf"
export OUTPUT="bands.out"

module load siesta/5.2.0.mpi

job-nanny mpirun -np $SLURM_NTASKS siesta < bands.fdf > bands.out

bands.fdf

# Siesta band structure calculation

SystemName          Si bands
SystemLabel         si_bands

# Use optimized geometry
MD.TypeOfRun        BS   # Band structure calculation

# Path in k-space
%block BandLines
  1  0.0 0.0 0.0  \Gamma
 20  0.5 0.0 0.0  X
 15  0.5 0.5 0.0  W
 10  0.0 0.0 0.0  \Gamma
 15  0.5 0.5 0.5  L
%endblock BandLines

# Include eigenvalues in output
WriteEigenvalues    T

# Rest of input (similar to previous)

Density of States 

submit_siesta_dos.sh

#!/bin/bash
#SBATCH -J siesta_dos
#SBATCH -N 2
#SBATCH --ntasks-per-node=28
#SBATCH -t 36:00:00
#SBATCH --mem-per-cpu=2G

export INPUT="dos.fdf"
export OUTPUT="dos.out"

module load siesta/5.2.0.mpi

job-nanny mpirun -np $SLURM_NTASKS siesta < dos.fdf > dos.out

dos.fdf

# Siesta DOS calculation

SystemName          Si DOS
SystemLabel         si_dos

# Projected DOS
ProjectedDensity    T
PDOS.Kgrid          10 10 10
PDOS.Emin           -20.0 eV
PDOS.Emax           10.0 eV
PDOS.Broaden        0.2 eV

# Rest of input (similar to previous)

Job Array para Varredura de Parâmetros 

submit_siesta_array.sh

#!/bin/bash
#SBATCH -J siesta_array
#SBATCH --array=1-5
#SBATCH -N 1
#SBATCH -n 1
#SBATCH -t 24:00:00
#SBATCH --mem=8G

MESH_VALUES=(100 150 200 250 300)
MESH=${MESH_VALUES[$SLURM_ARRAY_TASK_ID-1]}

export INPUT="template.fdf"
export OUTPUT="mesh_${MESH}/"

module load siesta/5.2.0.serial

mkdir -p mesh_${MESH}
cd mesh_${MESH}

# Modificar MeshCutoff
sed "s/MeshCutoff.*/MeshCutoff $MESH Ry/" ../template.fdf > input.fdf

job-nanny siesta < input.fdf > siesta.out

# Extrair energia total
grep "siesta: Total =" siesta.out | tail -1 >> ../energies.txt

Análise de Resultados 

analyze_siesta.sh

#!/bin/bash
#SBATCH -J analyze_siesta
#SBATCH -N 1
#SBATCH -n 1
#SBATCH -t 01:00:00
#SBATCH --mem=2G

# Extrair energia total
grep "siesta: Total =" *.out > energies.txt

# Extrair forças finais
grep "siesta: Atomic forces" -A 5 *.out > forces.txt

# Extrair geometria otimizada
for out in *.out; do
    echo "Geometria final para $out:" >> final_geometry.txt
    awk '/outcoor: Atomic coordinates/,/outcoor: End/' $out >> final_geometry.txt
    echo "" >> final_geometry.txt
done

# Extrair gaps (se houver)
grep "gap" *.out > gaps.txt

# Plotar com Python
cat > plot_energy.py << 'EOF'
import numpy as np
import matplotlib.pyplot as plt

data = np.loadtxt('energies.txt', usecols=(0,1))
mesh = data[:,0]
energy = data[:,1]

plt.figure()
plt.plot(mesh, energy, 'bo-')
plt.xlabel('MeshCutoff (Ry)')
plt.ylabel('Total Energy (eV)')
plt.title('Convergência de energia')
plt.grid(True)
plt.savefig('convergence.png')
EOF

python plot_energy.py

Referências 

Documentação: https://siesta-project.org/siesta/Documentation/
Tutorial: https://docs.siesta-project.org/projects/siesta/en/stable/tutorials/index.html
Pseudopotenciais: https://docs.siesta-project.org/projects/siesta/en/stable/reference/pseudopotentials/

Ver também

Quantum ESPRESSO - DFT com ondas planas
CP2K - DFT com bases mistas
Processando Simulações - Como submeter jobs