Differences between vasp5.4.4 and vasp6.2.1

[yuanchang_li]

Dear all：
I tried to use the vasp5.4.4 and vasp6.2.1 versions respectively to consider the U value（hubbard） and SOC effect, and then calculated the GW. After comparing the output file, I found that the response function and the micro dielectric matrix were normal, but there was a difference between the inverted dielectric matrix and the shielded coulomb potential. So I was very curious whether the version caused the difference? The following is the difference between the shielded Coulomb potential I compared. I thought it was the difference between the Intel linear algebra library and the Intel linear algebra library. After testing, I found that the reason was ruled out.
Best Wishes！
Quxiao

[martin.schlipf]

Could you provide the input files to your calculation please?

[yuanchang_li]

Dear martin:
Thanks for your reply!
The three files uploaded below are the reduced polarization function comparison file in vasp5.4.4 and vasp6.2.1(3.jpg), the INCAR file in the EXACT calculation, and the INCAR file in the GW calculation.
I found that it is normal in HEAD OF MICROSCOPIC DIELECTIRC SENSOR, but 1+v P, with REDUCIBLE POLARIZABILITY changes when calculating reducible polarizability. 6.2.1 is on the left and 5.4.4 is on the right. What I don't understand is whether relevant codes have been modified between the two versions?Please give me some help.
Looking forward to your reply!
Best wishes!
Quxiao

[martin.schlipf]

Sorry, but I cannot reproduce the calculations from the pictures. Please read the posting guidelines and provide the input and output of your calculation. VASP reports some information about the calculation in the header of the output files and when I have the raw files I can look at those. When the output looks reasonable I use the input files to reproduce your findings on our machines.