XunLei, most likely it's a cutoff issue. Years ago struggling for cutoff convergence was quite routine - there is even a paper by MCPayne and coworkers, in the early 90s on JPhysCondMatt, discussing how to correct approximately for too-small-a-cutoff.
Basically, as you enalarge your unit cell, your brillouin zone shrinks, and at constant cutoff it means that more lattice vectors (i.e. plane waves) enter into the fixed cutoff sphere. More plane waves means a systematically larger basis set (this is one of the good things of plane waves, it's easy to make the basis set more and more complete - a nightmare in Gaussian), and, variationally, a lower energy. You see in fact your energy drop going to the left. Check then what is the number of plane waves for each of your calculations, but most likely the wfc cutoff is the culprit. Your ecutrho corresponds to a dual of 12 (ecutrho=dual*ecutwfc), that is a very safe value, although higher values are occasionally needed for accurate magnetic properties (and lower values, down to 6, can be sometimes used). My rule of thumb is to start from 8, and repeat calculations with 6, 10, and 12, to figure out what works. (See e.g. the tests at: http://www.pwscf.org/pseudo/upfdetails.php?upf=C.pbe-rrkjus.UPF ). Note that if you do not have ultrasoft pseudopotentials in your calculations, a dual of 4 is the most you need (and, for super-large scale calculations, we used to play around with a dual of 3, again ages ago). This is due to the fact that if the wavefunctions are expanded up to gmax, the charge density (that is the square of the wavefunctions, in real space) will have in reciprocal space components up to 2 gmax (rho(r)=sum g=1,gmax, g'=1,gmax psi_g^start psi_g' exp i(g-g')r , and g-g' can then be anywhere between -2 gmax and +2 gmax) . The problem you see is particularly relevant in variable cell calculations (you do not want discontiuities while your cell breathes), and Sandro Scandolo (one of the fpmd authors, in quantum-espresso) has worked out a smearing scheme a few years ago (published, I think prb) that smooths away those disconituities. Happy computing, nicola XunLei Ding wrote: > Dear all, > I do a calculation to get the lattice constant of Ni crystal. > I scan the lattice constant from 6.60 to 6.70 a.u. using the bash file below. > The experimental value is 6.66 and is consistent with my result. > But it is clear that the curve departs into two parts at 6.66. > (see the picture at http://www.bsc.ustc.edu.cn/~dxl/download/Graph1.JPG) > Would you tell my why and how to solve this problem. > > Thank you! > > Ding Xunlei > > for a0 in 6.60 6.61 6.62 6.63 6.64 6.65 6.66 6.67 6.68 6.69 6.70 > do > cat >test.in<<! > &control > calculation='scf' > restart_mode='from_scratch', > pseudo_dir = '/home/bsc/dxl/pseudo/', > outdir='./' > prefix='zz' > tprnfor = .true., > for a0 in 6.64 > do > cat >test.in<<! > &control > calculation='scf' > for a0 in 6.65 > do > cat >test.in<<! > &control > calculation='scf' > restart_mode='from_scratch', > pseudo_dir = '/home/bsc/dxl/pseudo/', > outdir='./' > prefix='zz' > tprnfor = .true., > tstress = .true. > / > &system > ibrav=2, celldm(1) =$a0 > nat=1,ntyp=1, > nspin = 2, starting_magnetization(1)=0.7, > ecutwfc = 24.0, ecutrho = 288.0, > occupations='smearing', smearing='methfessel-paxton', degauss=0.01 > / > &electrons > electron_maxstep= 200 > diagonalization='' > conv_thr = 1.0e-8 > mixing_beta = 0.2 > &cell > cell_dynamics = 'none' > cell_factor = 1.2 > / > &ions > ion_dynamics = 'damp' > > / > ATOMIC_SPECIES > Ni 58.69 Ni.pbe-nd-rrkjus.UPF > ATOMIC_POSITIONS > Ni 0.00000000 0.000000000 0.000000000 1 1 1 > K_POINTS (automatic) > 13 13 13 1 1 1 > ! > > > > > > _______________________________________________ > Pw_forum mailing list > Pw_forum at pwscf.org > http://www.democritos.it/mailman/listinfo/pw_forum -- --------------------------------------------------------------------- Prof Nicola Marzari Department of Materials Science and Engineering 13-5066 MIT 77 Massachusetts Avenue Cambridge MA 02139-4307 USA tel 617.4522758 fax 617.2586534 marzari at mit.edu http://nnn.mit.edu
