Friday, December 17, 2010
shortcuts of openoffice are not working on slackware 13.1
Wednesday, December 15, 2010
scim on slackware 13.1
2. chmod +x /etc/profile.d/scim.*
3. create a script file scim.sh containing the followings lines:
#!/bin/bash
scim -f socket -c socket -d
4. chmod +x scim.sh
5. restart kde
Friday, November 26, 2010
Fedora 12 & Nvidia Video Card
Each time after updating the kernel, the Nvidia driver should be re-configured.
Press Ctrl-Alt-F2 to enter the command-line mode and login as root.
Reinstall the driver by running
./NVIDIA-Linux-x86_64-190.53-pkg2.run
and after that run
init 5
to enter the normal x11 mode.
Tuesday, July 20, 2010
Saturday, July 17, 2010
Experimental techniques
Friday, July 16, 2010
Graphene
Inside the codes of siesta & transiesta
siesta finds the density matrix by diagonalizing the Hamiltonian matrix, while transiesta makes use of the Green functions technique to obtain the density matrix.
siesta-3.0.rc2 is not working for transport calculation. I have not try if it works for band structure calculation.
I will use 2.0.2 version of siesta to do geometry optimization. After that use siesta-3.0b to do transport calculation.
Parallel compilation of siesta & transiesta
For siesta-2.0.2, run
./configure FC=gfortran –enable-mpi
make
For siesta-3.0, run
../Src/object_setup.sh
../Src/Configure FC=gfortran –enable-mpi
Sometimes the configure script can not find the location for libraries such lapack, blas, blacs, scalapack. The path of libraries should be specified explicitly as below:
BLAS_LIBS=-lblas
LAPACK_LIBS=-llapack
BLACS_LIBS=-L/usr/lib64/openmpi/lib -lmpiblacs -lmpiblacsCinit -lmpiblacsF77init
SCALAPACK_LIBS=-L/usr/lib64/scalapack-openmpi -lscalapack
Thursday, July 15, 2010
Inside the code
How to get the density matrix for siesta and transiesta? siesta finds the density matrix by diagonalizing the Hamiltonian matrix, while transiesta makes use of the Green functions technique to obtain the density matrix.
Settings about transiesta calculation
The ElectronicTemperature could be set to zero meV.
The maximum value of TS.ComplexContour.NLine is 11
TS.ComplexContour.Emin should be just a little bit lower than the smallest eigenvalues.