Quantum ESPRESSO Tutorial - SPIN POLARISED (MAGNETIC MOMENT) CALCULATION using BURAI

A detailed quantum espresso tutorial walking you through the process of performing spin-polarized calculations for magnetic systems and thus get the value of the magnetic moment per atom.
In the tutorial I use an Fe system as example, and accurately predict it's magnetic moment/atom.

[@PhysWhiz]

Actually, the video is a little flawed. That is, you don't need to run calculation for different starting_magnetizations and see which one gives the lowest energy. As you can see in the video the energy remains the same upto 6 decimal places which is probably because i set the scf convergence to only 6 decimal places. So essentially, all starting_magnetizations are giving the same energy. So in principle what one could do, is that they could try like a couple of starting_magnetizations such as one with all spins up or like 50% of the spins up, and if they both return almost the same energy then you have found your ground state magnetization state. If they are different then perhaps you have two local minima and you have to find like which starting_magnetization gives the lowest energy. But again if the difference is smaller then the scf convergence criteria, then they are all the same basically. Also, the concept of tot+magnetization is useful, when you already want know or want your system to have a particular magnetization. For example when running a calculation with a single O atom. The magnetization should be 2uB as there are two unpaired electrons in p orbitals. Even if you provide a starting magnetization you would get the final magnetization to be 2 uB. In light of this, to save the computaitonal time, you could just specify the tot_magnetization to be this value and calculate the correspopnding energy. However, Let's say you have an iron supercell (2x2x2). Then since you already know the magnetization from the unit cell calculation, you could in principle use that to specify the tot_magnetization, but if you're going t o be study the surfaces of doping etc. in iron crystals or supercells, then you cannot hardset the magnetization using tot_magnetization. As that would force your system to have that magnetization. So if you want to study magnetization in a system, then tot_magnetization is useless. In my video, rather than using different values of starting_magnetization and comparing the energies, I should have used different values of tot_magnetization, and then seen for which value the energy is minimum. But then again it will take a lot of time, so it's better to just use starting_magnetization to break the symmetry and it will automatically converge to the ground state magnetic state. Sorry for making this mistake in the video, but in principle, the mistake doesn't affect the result, it's just that the time spent in finding the lowest energy for diff starting_magnetizations is useless, as they all give the same energy.

[@johnconrad4439]

can you use norm conserving pseudopotential for spin polarized dft?

[@hilalkucuk5]

Thanks a lot. I learned many things from the video and your comments.Too many thanks :) No worries about mistake, I have learned other things from mistake too :) I tested using starting_magnetization starting from 0.1 to 1 by increasing by 0.1. I got similar total magnetisation with 0.2 and 0.3,0.4 .... until 1 but except 0.1. All of them gives similar results excepts one of 0.1. When starting_magnetization=0.1, total magnetisation gave totally different result. Is it expected case?

[@PhysWhiz]

@@hilalkucuk5 As I already mentioned in the above reply, you would have to look at the energies obtained using starting_magnetization=0.1 and other values. Whichever is the lowest, you could use that. Although since, all values except 0.1 converge to the same magnetisation state, I suspect that it might be the true ground state (global minimum).

[@hilalkucuk5]

@@PhysWhiz thanks a lot :)

[@heartking101]

Could please send me quantum express ?

[@bishalthapa4639]

much appreciated for the video, loved it

[@mudasirrashid4026]

Thankyou sir for such a valuable information.

[@MARahman-hd5iz]

Please make a video on magnetic properties calculation using CASTEP.

[@meghagoyal310]

Sir @Phys Whiz I want to perform SOC + magnetic calculations. And want to plot Spin up and spin down electronic band structure. Any suggestions Thanks in advance

[@hb8540]

Hello, I have a question regarding the non-collinear magnetism: For example if we would like to compare the magnetization in plane (IP) and out-of-plane (OOP) for a hexagonal structure, is it correct to run a non collinear calculation IP and OOP and compare the energies to guess the easy axis without any spin Orbit coupling (SOC), because I read somewhere that without SOC non-collinear calculation has no meaning, is that correct ? if so why ? thanks

[@RajendraSingh-bz2hu]

Thanks a lot. It is very useful to me.

[@zahidurrahaman428]

Thank You.

[@johnconrad4439]

I'm having trouble performing band calculations in spin polarized dft in burai. the following error always comes up: the bands code with constrained magnetization has not been tested

[@Mennu272]

Will we set starting magnetization only for the magnetic atom in the molecule?

[@Mennu272]

Please tell me when we have to calculate magnetic moment for a molecule like Mn doped SnTe, for 2*2*2 supercell, how can we check starting magnetization?

[@imdadkhan9631]

How we ll know about the crystal parameter mean lattice a b c o And atomic position

[@joynarayanmukherjee2128]

Can you show a noncollinear calculation with arbitrary angle of moments with respect to z and x axis? Is there any software to calculate those angles for a particular system? I am using VESTA where angle betwwen atoms can be determined easily but not the moment with axis.

[@priyambadaaiswarya1831]

Sir for ferromagnetic starting_magnetization1 &2 is all +ve & for AFM calculation starting_magnetization 1 is +ve & for 2 it's -ve For nonmagnetic calculation nspin=1 then what would be starting_magnetization value .

[@PhysWhiz]

For non-spin polarized calculations you don't need to provide any starting magnetization.

[@johnconrad4439]

Hello its me again. So im doing an optimization for Cl doped MoS2 and since i specified magnetization to be collinear it also calculates the total magnetic moment. Now in the iteration of the optimization when it converged, the total magnetic moment was around 0.67. However, after optimization is converged Burai/QE does one final SCF calculation. In the log it is titled: " A final SCF calculation at the relaxed structure" meaning in the iteration before this the structure was already relaxed. Now in this final SCF, the total magnetic moment is 0.03. Which value is right? And why is the value from the final scf very different if in the previous iteration it was already relaxed?

[@PhysWhiz]

This thread might be of use to you: lists.quantum-espresso.org/pipermail/developers/2017-February/001567.html

[@hilalkucuk5]

Dear Manas, thanks a lot for video and site. I wonder I am working on Cu embedded on graphene. Should I perform spin-polarized calculations on Cu embedded on graphene using scf calculation? Or Is just find spin-polarized calculations of Cu on crystal are enough? Thanks

[@ManasSharma07]

I'm a little confused regarding the question. But from what I understand, I think you should perform a spin-polarised calculation on the whole system as a whole.

[@mmekucuk]

Thanks a lot for very fast answer. I mean I embbed Cu on graphene and then check the lowest total energy vs total magnetic moment. I think it is correct way. After find lowest energy, should I use that tot_magnetisation or lowest starting_magnetization on input file?

[@hilalkucuk5]

@@ManasSharma07 Thanks again :) I still have a doubt on my mind. I will make a relax calculation. After finding starting_magnetization with lowest energy, should I put starting_magnetization on input file or tot_magnetisation to make relax calculation? The second question is you found 0.4 for starting_magnetization on video but you had put starting_magnetization=0.2 on input file. Did you put 0.2 for per atom?

[@PhysWhiz]

@@hilalkucuk5 Actually, the video is a little flawed. That is, you don't need to run calculation for different starting_magnetizations and see which one gives the lowest energy. As you can see in the video the energy remains the same upto 6 decimal places which is probably because i set the scf convergence to only 6 decimal places. So essentially, all starting_magnetizations are giving the same energy. So in principle what one could do, is that they could try like a couple of starting_magnetizations such as one with all spins up or like 50% of the spins up, and if they both return almost the same energy then you have found your ground state magnetization state. If they are different then perhaps you have two local minima and you have to find like which starting_magnetization gives the lowest energy. But again if the difference is smaller then the scf convergence criteria, then they are all the same basically. Also, the concept of tot+magnetization is useful, when you already want know or want your system to have a particular magnetization. For example when running a calculation with a single O atom. The magnetization should be 2uB as there are two unpaired electrons in p orbitals. Even if you provide a starting magnetization you would get the final magnetization to be 2 uB. In light of this, to save the computaitonal time, you could just specify the tot_magnetization to be this value and calculate the correspopnding energy. However, Let's say you have an iron supercell (2x2x2). Then since you already know the magnetization from the unit cell calculation, you could in principle use that to specify the tot_magnetization, but if you're going t o be study the surfaces of doping etc. in iron crystals or supercells, then you cannot hardset the magnetization using tot_magnetization. As that would force your system to have that magnetization. So if you want to study magnetization in a system, then tot_magnetization is useless. In my video, rather than using different values of starting_magnetization and comparing the energies, I should have used different values of tot_magnetization, and then seen for which value the energy is minimum. But then again it will take a lot of time, so it's better to just use starting_magnetization to break the symmetry and it will automatically converge to the ground state magnetic state. Sorry for making this mistake in the video, but in principle, the mistake doesn't affect the result, it's just that the time spent in finding the lowest energy for diff starting_magnetizations is useless, as they all give the same energy.

[@runClimbCode]

So even for an insulator that is AFM we must "set smearing"?

[@Bangalibabu95]

How to use electric field for Silicene?

[@Akshay-xo5iy]

Could you help me with Lithium Niobate?

[@science_10523]

nice and helpful video, but not able to run the program from Burai, only generating the file and running it in window cmd mode, which "exe" file I should initiate? plz upload one video for thermal property calculation

[@PhysWhiz]

You will need to use 'pw.exe' . Unfortunately I have no experience with thermal properties so I can't make videos on it anytime soon.

[@science_10523]

@@PhysWhiz thanks

[@priyambadaaiswarya1831]

Hello Sir, for ibrav=0 we have to assign cell parameters. Can u tell me where do I get these parameters?

[@PhysWhiz]

For a test calculation you can get information from the literature (experimental or theoretical studies). However, for a proper study you should optimize the structure using vc-relax option of QE in order to get the lowest energy cell volume and atomic positions.

[@gowthamv3687]

Can we do for alloy

[@johnconrad4439]

What are the specs of your computer? I tried doing spin polarised band structure calculations on 3x3 monolayer MoS2 (27 atoms) and its been 8 hrs already but its still not done :(. For reference, Im using an 8 processor 8 gb ram

[@johnconrad4439]

Btw sorry for asking these questions a lot. I just dont know who else to ask

[@PhysWhiz]

A monolayer calculation wiuld of course take time due to a lot of vacuum. But did you add vacuum or are you just using a 3x3x1 super cell. For monolayer there should be a large vacuum in the the third direction.

[@johnconrad4439]

@@PhysWhiz well, what i did was i started with an mos2 unit cell which is two layers of mos2. I made a slab in the 001 miller index and then i reduced the width of the cell until there is only one layer left.

[@dennisboakye5222]

@@johnconrad4439 What if you delete the atoms in the top layer? Will it still be a slab (monolayer)

[@yuhitgupta3388]

Sir I am getting spin up and spin down as u told in your video.... But I am getting only one band structure file Instead of two one for spin up and spin down... I have run the files Pw.x scf.out Pw.x nscf.out Pw.x band_1.out Bands.x bands.out As in bands.in file I am having only one output in bands.dat.... Please help in this regard. Please share ur email address if possible... Because I want to send u my files so that I can solve my problem....

[@nehakapila1330]

Is burai paid software?

[@PhysWhiz]

It's free and open source. You can check out my tutorials on it.