My Community

It used to be the case that NPAR must be set to the number of cores for hybrid DFT, GW, and RPA calculations. This is no longer needed for hybrid DFT a while ago. For GW calculations, I noticed that 5.3.5 and 5.4.1 do not complain if NPAR is not equal to the number of cores while 5.3.2 and versions before it would complain and stop if NPAR is not equal to the number of cores. However, the G0W0 results are different from those values calculated by commenting out NPAR and the calculation would hang starting from the second iteration for GW0 and NELM=4. Can anyone tell me what's going on? Shall I stick to the old way of GW0 calculation by commenting out NPAR?

Thanks,

huangj3

The use of NPAR is still under development.
HF and hybrid are OK, GW0 is under way.
We will announce when it is finished.
Until then stick to old regulations and
comment out NPAR.

Thank the admin for providing a quick response.

Best regards,

huangj3