On third step: I'm worry my system may back to it's initial configuration when no stress if I use the initial cell parameters. because the stress block are very small and in my case it looks like this: %block MD.TargetStress 0.000051 0.000002 0.000005 0.000002 0.000041 -0.000002 0.000005 -0.000002 -0.000784 %endblock MD.TargetStress
On Tuesday, August 19, 2014 10:56 AM, Максим Арсентьев <[email protected]> wrote: 1st calcul. usual str. relaxation 2nd change lattice param. by hand (5%) and fix cell (MD.VariableCell false), see output file for stress. 3rd: take stress values from output of 2nd to enter it to MD.TargetStress, use cell parameter without 5% decrease (initial) and MD.VariableCell true, see strain in output. Compare strain from output of 3rd with 5%. 2nd and 3rd need for check. For 50 % kpoint grid should be doubled, for 5% - increased about 5%. [100/(100+a)], where a - strain in % (negative value for decrease, positive for increase). Kpoint are in reciprocal space, so increase of the cell need decrease of kpoint and vice versa. 2014-08-19 19:49 GMT+04:00 Jaret Qi <[email protected]>: I have noticed the MD.TargetPressure and MD.TargetStress did not change significantly between step 2 and 3. I mean when make a strain calculation with MD.VariableCell true and false. And also I really do not understand why we need the third step i.e. that is when MD.VariableCell set to true, atomic positions and lattice vectors stay the same. Finally what is the criteria for changing the number of k points along the stress direction? So if I wish to make stress with 5%, 10%, 20%, 50% how much k points have to be increased? > >Sent from Yahoo Mail for iPhone > > > >________________________________ > From: Максим Арсентьев <[email protected]>; >To: [email protected] <[email protected]>; >Subject: Re: [SIESTA-L] how to process issues about strains(2) >Sent: Tue, Aug 19, 2014 11:02:44 AM > > > >No, all correct, I mean when doing structural optimization with >MD.TargetPressure and MD.TargetStress set MD.VariableCell need to be set true, >not false. >When doing str. opt. without strain leave MD.TargetPressure and >MD.TargetStress default (not mention them - you may comment these lines with >"#"). >I think you do not need to change k-grid and MeshCutoff (what is your k-grid?) >for only 5% change. If it would be e.g. twice smaller cell, then you need to >double your k-point in this direction with MeshCutoff the same maybe (maybe >somebody advice better) > > > >2014-08-19 12:47 GMT+04:00 邵德喜 <[email protected]>: > >I see.At first , my opinion "when I want to get he structure wuth 5% strain >in the z direction I think its cell is always fixed " is wrong. >>There are next two questions left: >>The first when I do twice structual optimization to get the structure without >>strain ,do I have to set MD.TargetPressure and MD.TargetStress both zero? >>The second :In the process to get the structure wuthout strain I've done >>convergence test .So I've got the corresponding value of k-grid and >>Meshcutoff,then when it comes to the structure wuth strain, >>do I need to do convergence test again?Or just keep k-grid and Meshcutoff >>same as that without strain? >> >> >>Thanks very much! >>Dexi Shao >> >> >> >>2014-08-19 1:40 GMT+08:00 Максим Арсентьев <[email protected]>: >> >> >>The third is MD.VariableCell true. >>>Correct, repeat again that before any full geom. optim. (MD.VariableCell >>>true) you should relax internal coordinates of your atoms (MD.VariableCell >>>false). >>> >>> >>> >>>2014-08-18 11:27 GMT+04:00 邵德喜 <[email protected]>: >>> >>> >>>I see.Let me give a summarization of it .Suppose we want to get the >>>structure wuth 5% strain. >>>>The first is to get the structure wuthout strain; >>>>The second change the lattice constant (also the atoms' coordinates) 5% >>>>,then do structural optimization with fixed shape of cell (MD.VariableCell >>>>false). After it get pressure and stress from the output file. >>>>The third do structural optimization another time ,but with the cell >>>>variable (MD.VariableCell false) together with the value of >>>>MD.TargetPressure and MD.TargetStress just as the second step shows. >>>>After the third step is done ,the structure witn 5% strain is obtained from >>>>the OUTPUT file. >>>>Is that all right? >>>> >>>> >>>> >>>>2014-08-17 21:15 GMT+08:00 Максим Арсентьев <[email protected]>: >>>> >>>> >>>>Yes, maybe first change lattice parameter by hand and fix shape of cell >>>>(MD.VariableCell false). You should obtain pressure and stress in your >>>>output. Then you can enter these values to MD.TargetPressure and MD.TargetStress with MD.VariableCell true (skip MD.RemoveIntramolecularPressure) to achieve your strain and to check this. >>>>>Remember that before any full geom. optim. (MD.VariableCell true) you >>>>>should relax internal coordinates of your atoms (MD.VariableCell false). >>>>> >>>>> >>>>> >>>>>2014-08-17 14:13 GMT+04:00 邵德喜 <[email protected]>: >>>>> >>>>> >>>>>Thanks very much! >>>>>>But what I consider is strain ,not stress.i.e. I noly know the change of >>>>>>the shape ,when it comes to where the change of the shape comes from ,I >>>>>>have no idea. >>>>>>By the way ,follow you if Ineed to set MD.RemoveIntramolecularPressure >>>>>>True and MD.VariableCell True to get the structure without strain ? >>>>>>Yours sincerely! >>>>>>Dexi Shao >>>>>> >>>>>> >>>>>> >>>>>>2014-08-17 17:17 GMT+08:00 Максим Арсентьев <[email protected]>: >>>>>> >>>>>> >>>>>> Dear Dexi! >>>>>>> >>>>>>> >>>>>>>There are variables which were created specially for these purposes - >>>>>>>MD.TargetPressure and MD.TargetStress >>>>>>> >>>>>>> >>>>>>> >>>>>>>2014-08-17 13:06 GMT+04:00 邵德喜 <[email protected]>: >>>>>>> >>>>>>> >>>>>>>Suppose we have got the structure without strain,then how to get the >>>>>>>structure with small strain? >>>>>>>>In my opinion,I should just change the LatticeConstant accords to the >>>>>>>>strain ,at the same time fractional coordinate is choosed.Then I do >>>>>>>>structure optimization with cell fixed.When the structure optimization >>>>>>>>is completed, the final structure is what I want (i.e. the structure >>>>>>>>with small strain ) >>>>>>>>Is that all right? >>>>>>>>Any comments will be appreciated. >>>>>>>>Dexi Shao >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>>-- >>>>>>> >>>>>>>Best wishes, >>>>>>>Maxim Arsent'ev, D.Sc.(Chemistry) >>>>>>>Laboratory of research of nanostructures >>>>>>>Institute of Silicate Chemistry of RAS >>>>>> >>>>> >>>>> >>>>> >>>>> >>>>>-- >>>>> >>>>>Best wishes, >>>>>Maxim Arsent'ev, D.Sc.(Chemistry) >>>>>Laboratory of research of nanostructures >>>>>Institute of Silicate Chemistry of RAS >>>> >>> >>> >>> >>> >>>-- >>> >>>Best wishes, >>>Maxim Arsent'ev, D.Sc.(Chemistry) >>>Laboratory of research of nanostructures >>>Institute of Silicate Chemistry of RAS >> > > > >-- > >Best wishes, >Maxim Arsent'ev, D.Sc.(Chemistry) >Laboratory of research of nanostructures >Institute of Silicate Chemistry of RAS -- Best wishes, Maxim Arsent'ev, D.Sc.(Chemistry) Laboratory of research of nanostructures Institute of Silicate Chemistry of RAS
