Error number of electrons and magnetization in OUTCAR
Posted: Tue Jan 22, 2013 1:14 pm
Dear all:
I have the same problem as the case in http://cms.mpi.univie.ac.at/vasp-forum/ ... php?3.7946, which lead to the error of number of electrons and magnetization in OUTCAR in my newest PC clusters.
For instance, I've calculated the graphene with 128 atoms which has number of electrons of 512.000000 and magnetization of 128.0000000. Therefore, the number of electrons and magnetization are 532.5587488 and 133.1396872 in the calculation, respectively, so that the calculated -1/2 Hartree and eigenvalues are wrong compared with using number of electrons of 512.000000 and magnetization of 128.0000000 (these two items were 512.000000 and 128.0000000 in the OUTCAR which is in my old clusters while using same input files to calculate).
The version of VASP our used in the newest clusters is 4.6.34, compilered by the last version of intel compiler and MKL (Xe_2013.1.1117); the MPI is using openmpi v1.6.3. (The version of VASP in old clusters is also 4.6.34, but used the intel compiler 11.1 and MKL 9.0, could it affect the results??) Incidentally, the CPU is Intel Xeon 6-core E5-2620 2GHz in my newest clusters, and the OS is using CentOS 6.0.
I've checked the Intel compiler whether it installed completely and it is OK in my system. Could everyone tell me how to solve this problem??
Thanks.
<span class='smallblacktext'>[ Edited ]</span>
I have the same problem as the case in http://cms.mpi.univie.ac.at/vasp-forum/ ... php?3.7946, which lead to the error of number of electrons and magnetization in OUTCAR in my newest PC clusters.
For instance, I've calculated the graphene with 128 atoms which has number of electrons of 512.000000 and magnetization of 128.0000000. Therefore, the number of electrons and magnetization are 532.5587488 and 133.1396872 in the calculation, respectively, so that the calculated -1/2 Hartree and eigenvalues are wrong compared with using number of electrons of 512.000000 and magnetization of 128.0000000 (these two items were 512.000000 and 128.0000000 in the OUTCAR which is in my old clusters while using same input files to calculate).
The version of VASP our used in the newest clusters is 4.6.34, compilered by the last version of intel compiler and MKL (Xe_2013.1.1117); the MPI is using openmpi v1.6.3. (The version of VASP in old clusters is also 4.6.34, but used the intel compiler 11.1 and MKL 9.0, could it affect the results??) Incidentally, the CPU is Intel Xeon 6-core E5-2620 2GHz in my newest clusters, and the OS is using CentOS 6.0.
I've checked the Intel compiler whether it installed completely and it is OK in my system. Could everyone tell me how to solve this problem??
Thanks.
<span class='smallblacktext'>[ Edited ]</span>