mfechner

excuse the delay here is my input see the reducebf ! tasks 0 maxscl 200 nwrite 50 epsengy 1.e-6 epspot 1.e-8 spinpol .true. spinorb .true. sppath '' ngridk 8 8 8 rgkmax 7.5 gmaxvr 18.0 lmaxapw 9 lmaxo 8 xctype 20 0 0 mixtype 1 reducebf 0.95 dft+u 1 1 : dftu, inpdftu 1 3 0.294 0.000 : is, l, U, J !! U on Tb-f, 0.294Ha = 8eV !rotavec ! 0 1 0 270 !! rotation of 270 degrees around +y makes a1 on +z avec -0.1716939763E-14 0.000000000 9.346584740 0.8584698813E-15 8.094379823 -4.673292370 -7.540006694 0.000000000...

two more small things to recommend: you should switch on reducebf, which decreases the applied magnetic field as you want to have at the end a computation without applied field. A value of 0.95 is good in most cases. In addition switching to linear mixing sometimes keep things also more stabe!

Dear Long, I tried your case and can indeed confirm the issue. However, could you check whether, when you rotate, there are any messages in the standard output files such as: "Warning(occupy): minimum eigenvalue less than minimum linearisation energy : -44.32280382 -4.383400000 for s.c. loop 137" This suggests that there is an issue when you rotate. Instead of rotating the structure using Elk, I tried directly using the rotated structure. (You can find it in GEOMETRY.OUT and directly copy it into...

Dear all, Is it possible to calculate the diamagnetic susceptibility of insulators within ELK by simply applying a finite magnetic field, then computing the LSJ expectation values, and finally taking the ratio of the total magnetization to the applied field to obtain χ? I recently tried this for diamond, but the χ value I obtained (~10⁻¹⁰) is much smaller than the experimental value, despite carefully checking all the units. Any thoughts on this? Thanks, Michael

Dear Mingsue, could I checked the symmetry of your calculation and got the following SHG tensor see attached png. So in principle the components xyy and yyx seems there to be forbidden so could it be that you looking at numerical noise. Can you check please the size of the allowed components and post them? best Michael

Dear Ming Su, could you provide more details here, what do you mean by not equal and which system (structure) do you calcululate? best wishes Michael

Dear Ming Su, could you provide more details here, what do you mean by not equal and which system (structure) do you calcululate? best wishes Michael

Dear Zhiwei, its tesseral spherical harmonic, so it is consistent with all other usage in the code. best wishes Michael PS: Remember that if you use the switch lmirep the columns are done with the adapted projections!