dear Jonathan, i have a question about your discussion, k-grid is reciprocal of real space, so if we have finite dimension in 2 dimension, and infinite length in z-direcrion, shouldn't we have large kgrid in x-y and small value for z, for example %block kgrid_Monkhorst_Pack 40 0 0 0.0 0 40 0 0.0 0 0 2 0.0 %endblock kgrid_Monkhorst_Pack
________________________________ From: 隆小江 <[email protected]> To: [email protected] Sent: Thursday, June 14, 2012 3:03 PM Subject: Re:Re: [SIESTA-L] about the K-point in transiesta Thanks, your response is greatly helpful to me! I appreciate it! At 2012-06-13 20:51:19,"Jonthan R. Owens" <[email protected]> wrote: Hello, > >The values for a, b, and c depend on your system. If, for example, your system is finite in the x and y direction, but infinite in the z-direction, then you will choose 1 k-point for the x- and y- directions, but (perhaps - it depends on the length) more in the z-direction. To figure out how many k-points you need, you have to do multiple runs and find the lowest number of k-points necessary to get an accurate energy value. It would also be instructive to read other published results that discuss the chosen number of k-points for calculations on systems similar to the one you're considering. > >TranSIESTA does not take c into account, because the boundary conditions for transport are different than they are for DFT. Specifically, you change from an infinitely periodic system in all directions, to semi-infinite electrodes in the transport direction, while the other two directions retain the same boundary conditions. The reason that changing c may change your results is because it does affect the SIESTA run - and therefore your DM, H, and S - which will then change your TranSIESTA results. > >Here is Monkhors and Pack's paper. Here is the TranSIESTA paper, which >discusses their chosen boundary conditions. > >Hope this was helpful, > >-Jonathan > >On 06/13/2012 07:48 AM, 隆小江 wrote: >Dear Siesta Users, >>I was confused about the K-point in transiesta calculation. for example: >>%block kgrid_Monkhorst_Pack >> a 0 0 0.0 >> 0 b 0 0.0 >> 0 0 c 0.0 >>%endblock kgrid_Monkhorst_Pack >>some articles set a=b=1, c about 20 to 30. >>while some others set a=b=3~6, c=1. >>I hear from the http://emuch.net/bbs/ that transiesta would not take c in to >>account, namely there would be no difference between c=1 and c=30. >>but I got different results when set a=b=2, c=1 and a=b=2, c=9. so I think >>that it might be incorrect. >>some other people says that the value of a, b, c should be different for the >>periodic systems and nonperiodic systems. >> >>I am wondering how to set them on earth. >> >>Thank you very much for your help. I look forward to your responses! >> >> >
