Pietro Bonfa

Dear Kay, thank you very much for your test! I managed to converge the DFT loop for my system with a couple of additional empty states, but I'm still very far from converging the total energy (and total energy differences) against this parameter (updated figure attached). One final question, if I may ask: when you mention the poor performance of SCAN for magnetism, are you quoting these papers https://journals.aps.org/prb/abstract/10.1103/PhysRevB.100.045126 https://journals.aps.org/prb/abstract/10.1103/PhysRevB.102.024407...

Dear Kay, thank you very much for your test! I managed to converge the DFT loop for my system with a couple of additional empty states, but I'm still very far from converging the total energy (and total energy differences) against this parameter (updated figure attached). One final question, if I may ask: when you mention the poor performance of SCAN for magnetism, are you quoting these papers https://journals.aps.org/prb/abstract/10.1103/PhysRevB.100.045126 https://journals.aps.org/prb/abstract/10.1103/PhysRevB.102.024407...

Dear Kay, thank you very much for the quick and extensive reply, as always! You asked many things so it took me a while to figure them out. Your input looks fine, except you may have to increase the number of empty states and check convergence. Indeed I had to, but this turned out to be very tricky! I managed to converge with up to 20 empty states, but couldn't go further. The (encouraging) result is attached. I tried hard to converge with 25 empty states but the nasty thing is that, depending on...

Dear All, I wanted perform MetaGGA simulations with plane waves, but before doing so I decided to test the accuracy of the results with Elk. While energy differences with PBE match quite well, Elk and plane waves (QE and VASP) predict a different ground state with rSCAN (see figure attached, error bars account for the use of different pseudopotentials). Do you see anything clearly wrong or suspicious with my input? tasks 0 metagga .true. ! convergence rgkmax 8.0 gmaxvr 14.0 ! default msmooth 4 xctype...

Dear All, I wanted perform MetaGGA simulations with plane waves, but before doing so I decided to test the accuracy of the results with Elk. While energy difference with PBE match quite well, Elk and plane waves (QE and VASP) predict a different ground state with rSCAN (see figure attached, error bars are due do the use of different pseudopotentials). Do you see anything clearly wrong or suspicious with my input? tasks 0 metagga .true. ! convergence rgkmax 8.0 gmaxvr 14.0 ! default msmooth 4 xctype...

Dear Kay, thank you very much for your prompt and extensive reply! Let me try to be 100% clear. When you say This is done in the routine 'gencore'. In the non-collinear case we have V^up(r) = V(r) + |B(r)| V^down(r) = V(r) - |B(r)| By doing this, you loose the directional information of B. Thus which is 'up' or 'down' is arbitrary. It also assumes that B is mostly in the same direction throughout the muffin-tin. That tells me that, when dealing with the core, the formalism is collinear therefore...

It took me about a year to find the time to go back to this topic, but I finally managed to dig a little bit more into hyperfine fields. I start from the end. You can see below a snippet from rhocore.f90 do ir=1,nri t1=abs(rhocr(ir,ias,1)-rhocr(ir,ias,2)) magmt(i,ias,:)=magmt(i,ias,:)+t1*v(:) i=i+lmmaxi end do where v is a unit vector that identifies the direction of the moment in the muffin tin. The question is: why there's an abs in the spin density originating from the core? I understand that...

It took me about a year to find the time to go back to this topic, but I finally managed to dig a little bit more into hyperfine fields. I start from the end. You can see below a snippet from rhocore.f90 do ir=1,nri t1=abs(rhocr(ir,ias,1)-rhocr(ir,ias,2)) magmt(i,ias,:)=magmt(i,ias,:)+t1*v(:) i=i+lmmaxi end do where v is a unit vector that identifies the direction of the moment in the muffin tin. The question is: why there's an abs in the spin density originating from the core? I understand that...