 There are some questions that people have raised in the live documents which are some of which are quite general and Maybe addressed if not now then Later Q&A today See we have we have one question already specifically about lecture, but I will hand over to Emiliano Dimitri Thank you. Thank you Arno. Okay, so I see from the chat the first question for from Merth that is say I cannot seem to understand the last part Of the lecture where you say this is not suitable for production run Can you elaborate on more on that? Merth is referring to the last slide of the Of the lesson where I give a short introduction about the basis set that The basis set let us say names that are available in CP2k when you want to decide which kind of Gaussian type orbital you want to use it In particular introduce the names like single zeta double zeta triple zeta and so on as a basis set with increasing accuracy which refers to the number of Of primitive fun or really contracted functions that are inside this basis set and that are necessary to describe the the molecular orbitals Usually The of course the first the minimal The minimal description that you can have to In order to have a Correct, let's say the wave function for For a molecular orbit is to include Let us say Include the functions only the function necessary to describe the occupied orbitals of the of the atoms involved in your Involving your molecular systems The orbitals in particular in for the basis set we are talking about here atomic orbitals because A molecule will be described as a Described the atoms will be described as a linear combination of the molecular Orbiter will be described as linear combination of atomic orbitals What is the minimal description of these occupied orbitals that we can get it using only just one Contractive function pair atomic orbital. This is called the single zeta basis set because Zeta is the The name that usually we give to the to the exponent in the In the primitive function in the Gaussian type orbital primitive function and So this means that I have only one Primitive of a contracting function for Atomic orbital to describe describe the atomic orbital which is occupied such Basis set allows you for sure to give a Correct description where correct means the minimal description possible For your molecular system correct, but it is not really sufficient to have a good accuracy a good the chemical accuracy in this sense and usually Let us say today's as a golden standard for sure then not golden standard is too much Let us say as a standard a minimal Minimal accuracy criteria to in order to publish a paper etc is not to just employ single zeta Basis set but at least goes to go one level after double zeta for example Ideally what you have do if you do a very good job. You should start to verify and do a convergent test of basis set increasing the step by step your Your basis set in terms of size of how in size of your of your basis set and see when The your calculation converge in the basis set so that you are sure that the the minimal one that give you a Nacuracy good enough is a sufficient is a the correct one that you can use for for all your calculations Related to this system. I would say that Of course, this is a very good practice to do when the system is small enough that allows you to get them, but Larger is the system more problematic is to do such a A test a convergent test that in any case would be the the best thing to do any time said that Said that when the convergent test is not really possible or is there limited to say at least the Criterium to use at least a double zeta size basis set is Is the one that currently is adopted and I suppose this is also the reason why in the tutorials we start using such a kind of size for the basis set just to give you Typical example of what we can And what we should do let us see Let me see so can I can it be used? For M and J opti but not on MD run especially for MD run. We need to choose this or a more accurate one I Would I would say what usually what I usually do etc. I usually usually Start with the single zeta Basis set just at the beginning to have an idea of my system, but also for geometry optimization After that I Immediately move after After testing that everything is working and so on and so on. I immediately moving to the double zeta once Really? The computational cost of course increase from this time But it's still usually at least for what I I have experienced so far is always affordable moving from single this zeta and double zeta and so After the preliminary test I move already with the double zeta Basis sets to better was it a basic set In any case Merth if you have if you want to elaborate more and ask me more just writing the chapter I Another question from Vinay. Can you please say differences between D3 dispersion and many body dispersion? Transistor state correction and that's the P2k have this many body dispersion correction as well Okay Many body dispersion. I do not know what you are referring to transition state correction Maybe you know you can elaborate. I can tell you something regarding the difference between DT2 and DT3 and Dispersion correction that were introduced in the in the in the lecture and Of course during the year we move from in this The FTD scheme we move from DT1 DT2 DT3 Making the correction the Functional form of the current order of correction. So the analytical form of the correction, but also the parametrization More accurate optimizing in some in this sense and the main Difference between the DFT D2 and DFT D3 is that the DFT D2 contain only two body Corrections with both body terms. Let's say terms that refer that that Taking to account only interaction between two body two atoms in the case of the DFT schemes while in the DFT D3 scheme there are there is only not only the term the Dispersion energy term that I mentioned in the in this live here the FTD in the as the DFT D2 scheme, but there are also additional terms in particular terms which Taking to account the interaction between three atoms at the same time and These of course are correction to the correction are terms which are Small with respect to the dominant effect of the two-body terms, but we have seen In the optimization while they tried and where they were trying to optimize the the parameter for to get the best The best dispassion correction that a three-body term was really essential to him to get an accuracy a Better accuracy in in this sense so Optimize only the two-body the parameter in the two-body Term was not if enough to improve the convergence. There's the sorry the accuracy Beyond a certain a certain threshold to make this is improvement the introduction of the three-body terms was crucial Did I understand Matt you say did I understand the Correctly from the presentation, but the hybrid Wait, I can add to the There is a different type of corrections Kachenka and chef of correction instead of Grime Not they know they are not available in 2k for now That type of one the dispersion corrections and Yeah, so I think for now in TP2k and promo TP2k interface you should stick to that if it is free I guess it's the best available method for now Of the dispersion correction so you can use Yeah, that's the second part of the question in a yeah. Thank you. Thank you. They make you Um, let's see Matthew didn't think what Lee They have the functions that they ever ubiquitous be three leap aren't recommended into P2k Why are not recommended? I Would not say that are not recommended a bit leap for sure what I can tell you be three leap Is a hybrid functional so as an exact exchange term is it? Inside that so of course the exact exchange term is computational more expensive really the They are more expensive Within the framework of the plane waves basis more of a plane wave as he sets and by the way this was also the reason why TP2k was Developed at a certain point historically the first code that implemented the Carparinello molecular dynamics Etc was a cpmd And from them that many developers move to other code Other code them in particular a cp2k with using this gpw approach this Combined plane plane waves and gaussian basis that approach In order to get the better performance in And the cure sum of deficiencies coming from an approach full plane wave like the cpmd one and one of these Let us say deficiencies was the fact that in cpmd that which is Full plane wave best set approach the Effort to calculate The exact exchange term is really large with cpmd is with cp2k this effort was Significantly in the radius of the still expensive, but it was significantly reduced and so With with cp2k The use of hybrid functional is possible is not only possible in the sense that is implemented, but because it's implemented or also in cpmd for example, but the computational of computational effort is Reasonable, let us say is more reasonable of course is not so straightforward like for example when you use gaussian, which is a full Localized basis set approach where the exact exchange and the effort to calculate this exchange is minimal But of course they still is still possible and B3 lip is a very popular hybrid functional because it was It been tested to be proved to be a Let us say a reasonable Functional in many kind of Situation approaches and a situation and the environment around of course there are alternatives The choice of the functional has to be done in general We have to pay attention when you choose it because because The results the calculations For your system could depend significantly from this choice Of course the first The first let us say way To move in order to make this decision should be going to the literature and look at what Was done at what was the results previously on system, which are very similar or identical to yours Because this is what is the first source any time going on the literature and look what the others already did and With more or less success and so you can have an idea the second when you have more or less an idea of which are the functional that better behaves with the with your system and with your Not only with your system, but also with the kind of calculation that you are interested in at this point Ideally you have to verify which ones which are functional are already Implemented on your code and for example CP2k in this sense as a very large offer of Offer of Functionals in general so you can Be rather confident that probably the best ones or some of the best that you have Identify in the literature for your system and your kind of calculation are available in CP2k do So about the hybrid functional sense P2k is a problem here That's the exact focus change is not very well working with plane waves. So for the exact focus change CP2k using direct Gaussian Direct Gaussian integrals like for example Gaussian So and and they also of course have a cutoff. So you cannot do it in periodic again. So what happens in your with your calculations that Hartree-Fock exactly change part of your Of your functional will not be periodic that the problem also it only have within the cutoff Calculates And that's also kind of the drawbacks. So you're losing partially the Periodist basically of your system. So it still can do periodic calculations, but Hartree-Fock part Exactly for the change will be not periodic So the dropback Yeah, but you can use them. So I mean, yeah, you can use them but with these drawbacks and also they will be slower slower Normal Cj Dj Okay, Thomas is asking I'm still confused with the non-local plane waves They seem very interesting for periodic system like crystal, correct Could you elaborate for their application in person for biomolecular system? Do we need to choose that them similarly as best set GTO? Are different type of non-local plane waves they why they do and they are not popular in other Quantum mechanics software at least I have seen them. Okay Okay Let's say why Say historically, let's start a bit historically. First of all, why we physicist Starting working with plane waves. You are right. For sure. The reason was because there were Solids solids in the first approximation you can consider as periodic one of the best way to treat it Condense matter solids, etc in this way is using the basis set, which is periodic as well And so plain ways starting this way Then moving towards Molecule, etc. Of course instead The first approaches was using basic set which are localized because molecule are molecules are Let us say very localized in space and so basic set which are localized Results to be much more convenient At At the time of course the most of calculation that were possible where only single point calculations a certain point people starting to become to became to become Interested to molecular dynamics to Abinacian molecular dynamics and Because going because they are interested to To investigate the dynamical behavior, not only the statistical behavior of the structure structural information of molecules All and large molecules and of course one of one of the first approaches was the born open-heimer scheme and Which would require? Let us say Which requires really a In optimization and a fun way function optimization at each time step so was a computational very expensive at the time and a breakthrough was in the 80s the The carparinello molecular dynamics the scheme of carparinello molecular dynamics that at that time Realists was able to speed up the The molecular dynamics quite a lot really you would be able to reach one order of money to the larger Time scale let us say and so starting to really see Events phenomena that were before impossible with with the computer At the time carparinello, however Requires as formulated as is requested plane waves a plane waves basic set And so for this reason they started to for this reason let us say they started to To be interested to use plane waves with molecules in order to be able to To be able to to do simulation with With the carparinello molecular dynamics schemes scheme And this was the reason why at the beginning it was used of course in the years the The meto that approaches to improve on one side the And on the other and both on the localize and the play and play wave ways one Of course now that wouldn't all this limitation In one sense on the other way are reduced The fact that CP2k use The combination of both localize and all and on the local basis set show you how I go it's improved in order to try to get the best from the feature of one and the others and And Still there are some feature of the localized the local the non-locum play waves waves which are Very very useful still when you deal with the molecules for example Still in the dynamical schemes When you have to write down the equation necessary to be implemented on a computer in order to In order to implement the a dynamical scheme You can see that if you use a localized basis set you have to take into account Additional terms to say the pull I for what I call pull I forces which are not present in When you write down the equation that you have to implement with the in a local basis set so this means that when You have to ask a computer to calculate your time step if you use a plane wave basis set Approach you have less term that have to be calculated Of course this end to be then you there are many magic that is around the numerical Implementation in in order to make them Let us say the calculation as fast as possible, but ideally consider that In the dynamical scheme non-locum plane waves a basic set is Simpler to as less term. Let us say to take into account with respect to localized one I want to just of course I Cannot enter now in them any detail, but I just just I want to just to give you an overview of which are the problem and the advantages of the two schemes and why people still are still are let us say Working and trying to optimize one and the other both with solid and more molecules even if the first Let us say Idea that you can have is that okay for more views localized basis set and for solids for condense matter I use plane waves basis set this is in the first approximation is true would be the best things to do but of course You can improve one on one side improve on the other side and what you want to have at the end is the scheme which is a As With performant as best as possible reaching the compromises in one sense of the other in particular when you are You are interested to the dynamical behavior because you want to do a vanish molecular dynamics Yes, so there is should be in terms of KMM what is important that you should Remember that I mean latest 80s and early 90s all molecular dynamics in classical have been done In non per is a in non periodic system or in periodic system is a cutoff So the problem is once the people realize that these produce a lot of artifacts on the edge For example, if you do a droplet simulation with cutoff without the periodicity Then you have a surface tension large surface tension for example, which can deform your system a lot in molecular classical molecular dynamics Then the people then in the 90s that starts using of first evolved summation. So fully periodic MD and then eventually the PME Algorithm evolved and then people start using full only fully periodic molecular dynamics in most cases Why because then you have these no problems with this surface tension effects and after that, of course Then when you want to move in such a scheme to the QMM then logically rises a question that we should also use some periodic scheme for QM forces as well and That's where there's this connection between the solid state and current biology because by physics Yeah, I can say because in in in in by physics as well. You are dealing usually with a system in a periodic box Of solvent. Yeah, basically set of solids periodic solid state You have no periodic box of solvent with inserted molecules and that's why you also want to do this periodic calculations and for this Gaussian basis set as Emiliano suggested there is no interesting periodicity With a localized motion basis, there is some tricks how you can impose that Even on the basis of the PME, but they're still evolving and they're still very numerically unstable And what yeah, so CP2k is one of the types how you can combine Localization from the localized basis set and predict of plain ways in one code. I don't know if there are any other codes which are doing similar things, but Yeah So that's why the idea is behind the CP2k. For sure CP2k is the more popular in this sense. Yes Or we can that employ this approach. There are others, but less popular in this moment There are other questions regarding the lesson Don't think so