Den ons. 6. maj 2020 kl. 22.01 skrev El-abed Haidar < ehai2...@uni.sydney.edu.au>:
> Yes that is exactly what I am saying. I just needed the confirmation. The > way I understood from gathering the previous siesta mails and after going > through the literature the following: > > > > in solid state physics: we need only 1 kpoint along a NON-periodic > direction. That is why, for a molecule or scattering region connected > between gold for example, only 1 kpoint is used. > > This is based on the fact that kpoints are a result of the Bloch theorem > which applies only to infinitely large crystals. With finite dimensions let > us say graphene, we have a crystal with periodicity in only 2 dimensions, > so only kpoints in 2 dimensions, and so on z axis is NOT infinite, it is > "semi" infinite, so we only use 1 kpoint still and as you said previous > studies Transiesta does a very different semi infinite DFT Hamiltonian > calculation yet a Hamiltonian unlike TbTrans which we aim to find the > current. > No, the z axis for graphene is not "semi" infinite. It is vacuum and hence not periodic (I have intentionally left out the discussion about the cell used in siesta). A semi-infinite region means it is infinite, but only in the negative or positive of 1 direction. In transiesta+tbtrans the semi-infinite region is replaced by self-energies that takes into account the infinite periodic part behind or in front of the scattering region electrode (semi-inf-dir -a3 or semi-inf-dir +a3, respectively). Hope that is correct and thank you again for your time. > > EL-abed > > Sent from Mail <https://go.microsoft.com/fwlink/?LinkId=550986> for > Windows 10 > > > > *From: *Nick Papior <nickpap...@gmail.com> > *Sent: *Tuesday, 5 May 2020 6:14 AM > *To: *siesta-l <siesta-l@uam.es> > *Subject: *Re: [SIESTA-L] Still confused about K points in siesta, > transiesta and tbtrans in siesta 4.1b4 > > > > Hi, > > > > Den søn. 3. maj 2020 kl. 22.03 skrev El-abed Haidar < > ehai2...@uni.sydney.edu.au>: > > Thank you for the compliment Nick, > > I was seeking an answer from yourself on why does the k points change from > 100 to 1 in tbtrans. > > The answer seems to be in the physics where the system is not being > periodic (semi periodic to be precise) therefore the kpoints from tbtrans > changes automatically to 1. But we use the 100 in the electrode calculation > and transiesta to find the Hamiltonian of both electrodes and scattering > region. Is that correct? > > In the electrode calculation you do use a high number of k-points along > the semiinfinite direction. > > Not in the scattering region. Please try and understand my question in 2. > > > > El-abed Haidar | Doctor of Philosophy (Science) > Condensed Matter Theory (CMT) Group > | School of Physics > THE UNIVERSITY OF SYDNEY | NSW | 2006 > > *From:* siesta-l-requ...@uam.es <siesta-l-requ...@uam.es> on behalf of > Nick Papior <nickpap...@gmail.com> > *Sent:* Saturday, 2 May 2020 7:07 PM > *To:* siesta-l <siesta-l@uam.es> > *Subject:* Re: [SIESTA-L] Still confused about K points in siesta, > transiesta and tbtrans in siesta 4.1b4 > > > > Hi, > > > > Great you have tried to fully understand this issue. > > > > Den man. 20. apr. 2020 kl. 22.01 skrev El-abed Haidar < > ehai2...@uni.sydney.edu.au>: > > Following up on that: In the manual it says that if we use* TBT kgrid > Monkhorst Pack (block):* > > This means it would act as the same as * kgrid Monkhorst Pack*, however, > this is used to have a separate k-sampling in the tbtrans utility. If it > does not exist it will use kgrid Monkhorst Pack > > But as i look into the output it seems to make it back to the default of 1 > 1 1. > > Not sure why? > > El-abed Haidar | Doctor of Philosophy (Science) > Condensed Matter Theory (CMT) Group > | School of Physics > THE UNIVERSITY OF SYDNEY | NSW | 2006 > > *From:* El-abed Haidar > *Sent:* Friday, 17 April 2020 2:44 AM > *To:* siesta-l@uam.es <siesta-l@uam.es> > *Subject:* Still confused about K points in siesta, transiesta and > tbtrans in siesta 4.1b4 > > > > Good afternoon to whom it may concern, > > i really want to understand the in depth behind the kpoints in electron > transport. I still feel after going through the siesta mail where I have > asked about it before in > https://www.mail-archive.com/siesta-l@uam.es/msg10916.html > <https://protect-au.mimecast.com/s/Gn7tCWLVXkUZlO9Rf6KvAP?domain=mail-archive.com> > leading > to https://www.mail-archive.com/siesta-l@uam.es/msg09386.html > <https://protect-au.mimecast.com/s/lgFTCXLW2mUjp25VuV3TvR?domain=mail-archive.com> > . > However, the answers were not clear to be honest as i did not get the > complete understanding out of this. I went back and read more about > periodicity as suggested back here in > https://www.mail-archive.com/siesta-l@uam.es/msg08489.html > <https://protect-au.mimecast.com/s/Cn8QCYW8NocmGM1EI9VR3m?domain=mail-archive.com> > and as Cubot mentioned its not about periodicity that is usually defined in > solid states.* Any resources on k points for transport we can rely on or > tutorials? Because* > > I am more confused on the k points when defined for electron transport > study. So im trying really and really would appreciate if there is a > tutorial in the future for the latest siesta version to how we use k points > BECAUSE > > > > So far i understand that we will go through 3 steps > > 1-transiesta <electrod.fdf >electrode.out to calculate Hamiltonian and > overlap matrix of electrodes > > 2-transiesta <scat.fdf >scattering.out to calculate density matrix of > scattering region ] > > 3-tbtran <scat.fdf >.tbtrans.out to calculate current and transmission > > When papers define the k points to be 1 1 100 along electron transport i > would assume we use in all siesta versions %block Monkhorst Pack and > through out the 3 steps i assume i am using 1 1 100 in my entire > calculation. *I found out recently that once siesta run is done in step > 2 and starts a transiesta run the k points change suddenly to 1 1 1. > WHY???? It was > mentioned https://answers.launchpad.net/siesta/+question/686603 > <https://protect-au.mimecast.com/s/2qP1CZY1NqiAOl4rcy0zXs?domain=answers.launchpad.net> > but > not enough explanation on the reason why. * > > > > I am not sure why especially that i want my study to be infinite along x > and y ( or lattice vector a and b) and finite in z ( or c vector). > > *Now if I am suppose to add a block in each step above mentioned, what > should I do to make sure all 3 steps have 1 1 100? IF NOT AGAIN WHY?* > > *I fully understand that i should do convergence tests for k points but i > need to know also if that is always the case. Because i read that 1 1 100 > is more than enough (based on literature and reading around the siesta > mail)* > > > > *Would anyone be able to give their in sight thoughts as detailed as > possible. Thank you* > > > > I would suggest you to first conceptually figure out what the NEGF code > does. As per your 3 steps above > > 1. transiesta elec.fdf > > This calculates a _truly_ bulk electrode. Ask your selve this: > > a) how do you approach the truly bulk charge density in a regular > calculation? Which parameter should be *tuned* to find the truly bulk > electrode? > > b) You want the electrode to be added to a device region. In the device > calculation you have a particular direction of the electrodes that are > "semi-infinite". You want the two systems to be commensurate in terms of > calculation parameters. How does this relate to a)? What does this imply > for the semi-infinite direction, and for the transverse directions? > > 2. transiesta scat.fdf > > Here you calculate the device region where you place your _truly_ bulk > electrodes. What does the NEGF implementation state about electrodes? What > does an electrode mean in a device region system? And more importantly, if > you have 2 electrodes both with the same semi-infinite direction. What does > this mean for the k-point sampling along that direction? Why shouldn't it > matter whether you use any number of k-points? (once you figure > out/understand this it means you understand the NEGF concept ;) ) > > 3. tbtrans scat.fdf > > This calculates the spectroscopic quantities, such as DOS. > > There is no conceptual difference from TBtrans and from a regular PDOS > calculation using Siesta. > > E.g. if you use a k-point sampling [nx=3 ny=3 nz=3] in siesta, would that > be enough to get a good PDOS estimate? > > If not, why? And how do you resolve this? > > > > El-abed Haidar | Doctor of Philosophy (Science) > Condensed Matter Theory (CMT) Group > | School of Physics > THE UNIVERSITY OF SYDNEY | NSW | 2006 > > > -- > SIESTA is supported by the Spanish Research Agency (AEI) and by the > European H2020 MaX Centre of Excellence (http://www.max-centre.eu/ > <https://protect-au.mimecast.com/s/t6kFC1WLPxcxv3z2IYR87X?domain=max-centre.eu/> > ) > > > > > -- > > Kind regards Nick > > > -- > SIESTA is supported by the Spanish Research Agency (AEI) and by the > European H2020 MaX Centre of Excellence (http://www.max-centre.eu/ > <https://protect-au.mimecast.com/s/t6kFC1WLPxcxv3z2IYR87X?domain=max-centre.eu/> > ) > > > > > -- > > Kind regards Nick > > > > -- > SIESTA is supported by the Spanish Research Agency (AEI) and by the > European H2020 MaX Centre of Excellence (http://www.max-centre.eu/) > -- Kind regards Nick
-- SIESTA is supported by the Spanish Research Agency (AEI) and by the European H2020 MaX Centre of Excellence (http://www.max-centre.eu/)