ABINIT Discussion Forums

Hi, I was having some issues with SCF cycle convergence. Therefore, I set iprcel value to 45. But this in fact increase the memory requirements by multiple folds and the job aborted with out of memory error. Can anyone suggest if there is any way to reduce the memory requirement and get the SCF cycl...

Thank you.

Hi All, I am studying a 4x4x4 supercell of Silicon with one palladium atom and one H2 molecule. Could anyone suggest suitable diemac value? There are 340 Silicon atoms in my system. I know suggested diemac value for Si is 12. But since Pd is metal and metal needs high diemac value, I am not sure abo...

Professor Zwanziger,

Thank you for your reply. I already figured out the parallelization techniques. I run those in a supercomputer facility with access to hundreds of cores.

Regards,
Mawla

Thank you for your suggestions. I have been tweaking the npkpt, npband, and npfft parameter values to reduce the memory needs. But I was wondering if there exists any straightforward rule or it's always such trial and error basis?

Hi, I am doing some calculations and unfortunately it exceeded the memory. To solve this issue, I set the mkmem variable equals to the number of k points in the input file as per the suggestion in the log file. But apparently, I am receiving following error: --- !ERROR src_file: chkvars.F90 src_line...

Thank you so much. It totally works!

Hi, Could anyone let me know if it is possible to relax only part of a structure? For example, I have a 4x4x4 Si supercell with 1/4th of a pore (diameter=3 nm) in a corner. I need to relax only the Si atoms near the pore. Is it possible in abinit? I have all the atom positions in Angstrom unit. Than...

Hi, I am a new abinit user. I have 4 Silicon atoms and 1 Pd atom forming a tetrahedron. I am trying to figure out the bond length between Si & Pd by varying the distance between 2 atoms (1 Si & 1 Pd) only out of all 5 atoms. Is it possible? I know if there were only 2 atoms (1 Si and 1 Pd), ...

Hi, I am a new abinit user. I am trying to calculate total energy of silicon supercell containing 4 conventional unit cells in x,y,z directions and 1/4 th of a pore (pore radius = 15 Angstrom). I ran the simulation in available supercomputer facility at the university but it ran out of wall time aft...