I have calculated the magnetic moment of Nd2Fe14B (P42/mnm ) with PBE successfully from wien2k 19.1. Then I substituted Ti、Cr、Mn、Co、Ni、Al in six crystallographically inequivalent Fe sites respectively to obtain Nd8(Fe55X1)B4,( X=Ti、Cr、Mn、Co、Ni、Al). I calculated these the magnetic moment of these new systems with PBE successfully as well. Similarly,I substituted La、Ce、Pr、Sm、Dy in 2 crystallographically inequivalent Nd sites respectively to obtain (Nd7Y1)Fe56B4,( Y=La、Ce、Pr、Sm、Dy).The calculations are smooth as well. However, when I substitute elements in both Nd and Fe sites to obtain (Nd7Y1)(Fe55X1)B4,(X=Ti、Cr、Mn、Co、Ni、Al,Y=La、Ce、Pr、Sm、Dy).The calculation result is not convergenced,neither show a convergence trend after a iteration of 30.I have followed steps below to calculate the magnetic moment of the system.
1) Initialization has been done in batch mode (with init_lapw) 2) init_so 3) init_orb -orb -f -c In these steps,I follow the hints to set the parameters.The steps run without error but the results didn't reach the convergence.To be more specific, it doesn't show a convergence trend.I checked the structure files and I am unable to find whats going wrong.
The convergence problem is an usual problem in magnetic calculations and to resolve such problem try to use the FSM procedure which has 2 functions: 1- resolving the convergence problems and looking for a second magnetic solution as in my talk
https://aljazairi1970.blogspot.com/2021/02/le-workshop-de-mostaganem-la-procedure.html
good luck
In my opinion, when the SCF calculations is made with spin polarised configuration then it is always important to give the charge convergence criteria. The magnetic calculations sometimes converge into local minima, i.e. the energy convergence criteria satisfied but magnetic ground states did not reach. The one possible way to converge the calculation is to increase the number of iteration and give the charge convergence while running the SCF calculation by runsp_lapw -cc 0.001 (or 0.0001), and see the result. Another way is to do the fixed moment calculation, please check the user's pdf file or some example. If your system has an anisotropic effect, then also magnetic moment will be sensitive with the site adopted for the dopant element. It is required to be careful check. Let me know if you still have issue, I can check it and help.
Anqi Zhang
Ill-conditioned (ill-posed) problems often cannot get a better solution by ' increasing the number of iterations'. With certain mathematical functions, expressed as a sum of a series, the error terms INCREASE after a number of iterations and divergence (not convergence) occurs. In these cases it is vital to stop the calculations at the optimum number of terms. Alternatives here, once the problem is recognized, is to approach the problem via a re-formulation of the terms/boundary conditions and confirming that the generated solutions agree.
Abderrahmane Reggad Thanks a lot for your answer! As i checked the guidebook, FSM method is to constrain the total spin magnetic moment per unit cell to a fixed value.But since what i want to get is the total magnetic moment, how can i apply a unknown value in the problem?
Saurabh Singh
H i，thanks a lot for your help! I didn't checked out proper fixed moment calculation example via the web, would you please talk more in details?
Dear Anqi Zhang,
Please follow the discussions made in wien2k forum using below link.
https://www.mail-archive.com/[email protected]/msg10426.html
In case you need the assistantship in calculation please feel free to contact me. I will try my best to help you.
Good luck.
Anqi Zhang
There is another purpose of the FSM procedure. try to check the page 33 of this file http://susi.theochem.tuwien.ac.at/reg_user/textbooks/WIEN2k_lecture-notes_2013/magnetism.pdf
and this http://wien2k-algerien1970.blogspot.com/2019/11/wien-fixed-spin-moment-calculations.html
- Badges
- Science topic