 So do you see my terminal now? Yes, we can see it. Basically, if you go to each directory, you can find this file. Lead me. If you just follow this instruction, then you can just perform the tutorial. For example, in the case of Garlin-Arsonite case, you should open the readme file. There is an instruction. So we first perform SCF and NSCF calculations. If you want, can we use parallelization? Or if you want, you can also parallelize pw.x? I think up to five MPI processes should be okay on the virtual machine. But it is not expensive. The single parallelization is enough, I think. And then move on to the symmetry-adapted one-year function. In this case, SP model centered at the Arsonite site. Apologies, it's Friday. No, it's not five. It's four MPI processes, my bad. You can also look at, in this directory, you can also find this SAWF tutorial instructions in the directory. You can also have a look at this instruction. This is created by Barrio. And I'd like to thank Barrio for creating these very nice instructions. Basically, the instructions are very same as the readme. So you can either look at readme or this instruction. And as I already said, if you go to each directory, we have the REF reference directory. And if you look at the reference directory, you can find the input file, the pw21.in, and also the w190.in, and also the reference output file, wout. So if you do not have confidence in the input file or the output, you can refer to these files. And also in the case of Garium Arsonite, or the Kappa case, in the original paper by Ray Sakuma in 2013, so in the original paper, considers Garium Arsonite. And the Kappa as an example. So you can also compare your result with the original paper. You can also find this paper in the github directory. So in the case of Garium Arsonite case, the maximum operator is 1A functions, the sp3.1A orbitals. So it would be interesting to also work on these maximally-operated 1A functions. So in this case, the 1A orbital center at Arsonite site just, it is like a subtle point. So if we just put a small perturbation to the initial, the position of the initial projection, then it quickly flows into maximally-operated 1A functions. Then the total spread of the maximally-operated 1A function should be smaller than the symmetry-adapted 1A functions. Excuse me. Can I ask one question? Sure, sure. For this tutorial, Garium Arsonite, the valence band, is it possible to quickly check if the top of the valence band has the right degeneracy? That created 1A functions? Yes. So if you plot the band, or you can, I think. So in the GNU, the Garium Arsonite underscore band GNU, so this would just plot this? Yeah, yeah, yeah. I think you can plot the band structure using this. And then just, we can zoom in and see. Thank you. Most probably, if the symmetry of the 1A function has the proper symmetry, then also the band structure should also have the proper symmetry. So you can also check the symmetry by the standard output. In this case, all the three orbitals are equilibrium, and it's actually at the atom center, so the 1A band should also satisfy the symmetry. Hello, can I ask you a question? Hello? Sure. Yeah, so in the case of a material where the band minima is not at a high symmetry point, for example, silicon, the conduction band minimum is between gamma and x. In that case, if you use maximally localized 1A function and you are increasing the grid, the k point grid, then the minima of the conduction band will move slightly. If you are using symmetry adapted 1A function and you are along this high symmetry line, gamma x, will this help? So if you then change the coarse k point grid, will the minimum of the conduction band move or will it stay the same? Well, if, so, basically in the case of maximally localized 1A function and the symmetry adapted 1A function, if the number of the 1A orbital is the same as the number of bonds, it basically, both 1A function, just reproduce the DFT band structure, but DFT band structure, right? So this band minima program is already exists in the case of DFT band structure or just the program of the maximally localized 1A function? It's a problem of the maximally localized 1A function. So somehow you don't exactly reproduce the conduction band minimum and the band will slightly change, move, if you change the k point grid. So it will converge, but it will slightly move, converge slowly. So for very... So you mean that if the number of k point is small or the minimum? Yeah, 666 or something like this and then 888, 1010, then you will converge but it will move. So it's a slow convert. And then it does not help. It doesn't help. Okay, okay, thank you. This is because the hopping structure, hopping is rather long range. So basically the spread of the symmetry adapted 1A function should be the same or larger than the maximally localized 1A functions. Then the hopping range is basically the same or even longer than the... Okay, thank you. Thank you. I don't know, I don't know. But this problem comes from the long range hopping. So I don't think it helps, but if this long range hopping is not allowed by symmetry and it is from just garbage garbage of maximally localized 1A function, namely it is not essential, then it might help. But my opinion is maybe it does not help. To be real, what you, when you said small perturbation around what number I'm... Okay, so in the example of the maximally localized 1A function in the Garin-Arsonite case, so in the reference input, I put this number. So 0.25 is the position of arsenic site, but I put 0.251, so 0.001 perturbation to the projection. It can be smaller, but maybe it need more iteration to clearly break the symmetry. So if you have big perturbation, then it quickly flows into maximally localized point because it is a subtle point. I have a question. In the copper example, we just copy the same file from Garin-Arsonite. Yeah, but if there are some utilities that generates this file for a particular set of symmetry or just from giving a few generators of the symmetry group, I mean something similar that can generate if I observe the need of typing all this. Okay, so now we have a question on copper. So let me explain this. Yeah, yeah, I'll keep the guy on arsenic, I jumped on copper. Yeah, the reason why we take from the Garin-Arsonite is that the full symmetry from copper, then we remove the inversion symmetry, then the set of symmetry operation becomes the same as the same as that of Garin-Arsonite. The reason why we... No, I understand, yeah, I understand it. Just the question that can we generate this file somehow without running calculation for Garin-Arsonite? From PW, I see from PW, we cannot customize symmetry, right? If they detect the full symmetry, if we use symmetry, then this symmetry adopted mode just use this information from that. In principle, if it is not easy to write some program to remove the inversion symmetry from this full symmetry group, but for the moment, if we want to customize the symmetry, then you need to create the dot sim file by yourself. And basically, the... The style of the sim file is very simple, just... No, I know, I'm just asking. For the moment, you need to create the files. So in the case of, so now we have a question on Kappa. So in the special case, actually we can also customize the symmetry operations, what we use. And if we want to customize the symmetry to be used in the symmetry adopted mode, you can put the additional input file, additional input, read sim in the PW2.1 file. Then in this case, this PW2.1A requires the dot sim file. Then you can customize the symmetry operations to be used in the symmetry adopted mode. But in the case of, in the example of this Kappa tutorial, we try to create the S orbital centered at one fourth, one fourth and one fourth. In this case, the symmetry of this projection are lower than the original Kappa symmetry. Because here, the existence of this S orbital breaks the inversion symmetry. To keep the inversion symmetry, we need additional input like minus one fourth, minus one fourth, minus one fourth, minus one fourth. Then if we create seven orbital, one orbital, then you can keep the original full symmetry. But here we omit this S orbital at minus one fourth, minus one fourth and minus one fourth. Then with this set, we do not have inversion symmetry. Then in this, the symmetry also need to be compatible with this initial projection set. Then you can customize the symmetry file. Then as I already replied to the question, for the moment, if you want to customize the symmetry, you need to create the dot sim file by yourself. But in this case, the symmetry of the Kappa without inversion symmetry is exactly the same as the symmetry of Garry-Marsenaid. So then in this case, you can just copy from the Garry-Marsenaid. Can I just add that nice website where you can look at the symmetries of different crystal structure, is the Bi-Bao crystallographic server where you can look at all the weak of positions of all possible groups? I see, so actually I have a question. The question is, can you clarify why you set one fourth, one fourth, one fourth S-Obital or in the Kappa example? Why do we want to let S-Obital located at one fourth, one fourth, one fourth? And actually this is a kind of, this example is a kind of artificial example. We can use physically there is not so much much motivation to do that actually. But just for demonstration actually, we can customize the usage. Just for demonstrating the custom usage, we put this projection, which is this projection. But is there actually some example where it is really useful to customize it? To use some reduced symmetry for the Vanilla functions? I don't think there's not so much such situation because it means that if we want to lower the symmetry from the original full symmetry, right? So basically what we want the symmetry of the function basically physically, it should keep the symmetry of the original compound. Only if you want to lower the symmetry of the function, you customize this kind. Can it work in the opposite way that you actually make a higher symmetry? Hiya. For example, yeah, so that say your system breaks some system but maybe only slightly due to some pattern displacement. And you far put an extra symmetry. What will happen? Yeah, symmetry, I have never thought about that. In that case, because the DFT panel structure, it breaks the symmetry. So, for example, if the symmetry is lowered at some K point, the degeneracy is lifted. But if we use higher symmetry, it means that if we cannot reproduce this lifting of the Van. Yeah, of course, but what will happen? Just the other question. What will happen? Yeah, I think it will crash or will it just restore the degeneracy? I think it crashed because it used the original, this D-tuder matrix is how we transform the, how the broccoli function is transformed. And this D-tuder and this capital D is not compatible in that case. This, in this D-tuder case, the degeneracy is lifted but in this capital D, there is a symmetry. So, it is incompatible. So, I think it produced an error but I have never tried that. Okay. I have another question. In the case of Kappa 2D-like running functions of the same spread, while the other 3D-like running function also have the same spread but in a different part. Yeah, it is due to the, it is the EGT 2G spread. Okay, so we have one question. I want to ask you. Okay, no question. Hello. Yeah, I have a question from the Kappa example. There, I ran the calculations to check whether the spread of the one-year functions are similar to what I am getting from the reference you have provided. For example, omega i, I am getting is 4.097 and in the example or in the paper, it is given 3.968. So, is it because we have used some approximations so it is not the ideal case? Therefore, this slide discrepancy is coming between my calculation and, yeah, exactly what you are showing that I also got those results, yeah. In the reference, in this file? Yes, so in the output file, if you go, cu.waut, yes, exactly. Omega i, 4.09, and what you get is? Yeah, and if you compare with the paper, it is... A paper, sorry, the paper, first they use the different pseudo-potential and the difference might come from the different pseudo-potential. Okay, thanks. Also, the K point. K point I think is the same. The energy beam, another reason for that is the K grid and also the K grid should be the same but also the energy beam or something like that. But in principle, they use the different pseudo-potential and I don't think they use quantum espresso called the DFT structure calculation so the difference should also come from that. Okay, thank you. I have another question. Question in the chat. That is the origin of this. What? I mean, it has the FCC in the copper, I think it is FCC structure, right? In this example. Then in some orbital, the lobo of the sum orbital points to the other copper orbital but other orbitals are not. Then this gives the splitting between EG and T-solution even in this single copper case. The position of the copper atoms breaks the spherical geometry. Hi, I have a question about the dot sim file. So if I understood correctly, in the case of copper, we have to remove inversion symmetry because in one of the two positions where we want to get the vanishing functions then the inversion symmetry is not contained. But then on the other bike of position, it is contained. So are we losing information of the set symmetry group of the other bike of positions? Yeah, that's right. For the orbital, that's right. So it's not possible to specify different set symmetry groups for the different bike of positions? It might be possible, but for the current implementation, that kind of use is not supported. Okay, thank you. Can you hear me? Yes. This is Sophie. Thanks for the nice talk. I have a question about the, actually the NSCF calculation, you know, that one has to do. I once tried this for a distorted perovskite where you have like four symmetry equivalent sites. And there, if I, if I don't specify, no sim equals true, then quantum espresso will add additional k points to the one specified in order to preserve the symmetry of the grid somehow. Did you ever encounter this? So, you know, because of that, I wasn't able to run the, to run the symmetry adaptive mode. So which compound are you using? It's like, it was calcium VO3, but like any perovskite structure? So let me discuss data in more detail. So if you send some input file or to just send email to me, I can check. Great. Yeah, I will do that. Thank you. I think in a few minutes, the tutorial session will finish. So if you have questions, please ask me now. Any questions? Any questions from the Adriatica? Okay. There is one question here. Is the symmetry adapted option available when we consider Vanier 90s library in Avonito code? So again, available for... I think the question was, if one can use the symmetry adapted mode in the symmetry adapted method with the Vanier library mode. Vanier library mode? Ah, for the moment, no. I think it is only implemented. I forgot, but I think it is only in the standalone mode. Yeah, I suspect it must be so, because you know, the library mode contains, at the moment, only the basic features. While actually next week, we will have developers meeting where there will be discussions on some work that has been going on in the UK to sort of restructure the library mode to contain. In a way that would contain all the features of Vanier 90, but this is not there yet. So I suspect that most of the specific features are not there. Yeah, thank you. Thank you for the clarification. Okay, I think we should stop here. Let's thank our speaker again.