Re: [Wien] strange(?) LOPW - Error
Thank you Ciao Gerhard DEEP THOUGHT in D. Adams; Hitchhikers Guide to the Galaxy: "I think the problem, to be quite honest with you, is that you have never actually known what the question is." Dr. Gerhard H. Fecher Institut of Physics Johannes Gutenberg - University 55099 Mainz Von: Wien [wien-boun...@zeus.theochem.tuwien.ac.at] im Auftrag von Peter Blaha [peter.bl...@tuwien.ac.at] Gesendet: Montag, 29. Januar 2024 09:38 An: wien@zeus.theochem.tuwien.ac.at Betreff: Re: [Wien] strange(?) LOPW - Error I crosschecked lopw.f again. It has a remark about an "experimental check", that not only the Overlap matrix elements including phase factors are orthogonal, but already the K-vectors are not linear dependent. The latter is a much faster check, but maybe too restrictive. When I remove this additional check, it runs through and the eigenvalues look reasonable (and for k-vectors which run through, the eigenvalues are identical (although the K attachments are different). You may want to test this version. Regards Peter Am 28.01.2024 um 12:20 schrieb Fecher, Gerhard: -- -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-165300 Email: peter.bl...@tuwien.ac.atWIEN2k: http://www.wien2k.at WWW: http://www.imc.tuwien.ac.at - ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
Re: [Wien] strange(?) LOPW - Error
I crosschecked lopw.f again. It has a remark about an "experimental check", that not only the Overlap matrix elements including phase factors are orthogonal, but already the K-vectors are not linear dependent. The latter is a much faster check, but maybe too restrictive. When I remove this additional check, it runs through and the eigenvalues look reasonable (and for k-vectors which run through, the eigenvalues are identical (although the K attachments are different). You may want to test this version. Regards Peter Am 28.01.2024 um 12:20 schrieb Fecher, Gerhard: -- -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-165300 Email: peter.bl...@tuwien.ac.atWIEN2k: http://www.wien2k.at WWW: http://www.imc.tuwien.ac.at - SUBROUTINE LOPW(NAT) ! use matrices, only: HSROWS, KZZ, XK, YK, ZK use lolog, only : nlo, ilo use lstapw, only : NV use rotmat, only: ROTIJ, ROTLOC use struk, only : POS, MULT, NDF use parallel, only: myid,abort_parallel IMPLICIT NONE INCLUDE 'param.inc' ! !Scalar Arguments ! INTEGERNAT ! ! .. ! !generates the LAPW (K+G)-vector for local orbitals ! ! .. ! !Locals ! INTEGERIA1, IEQ, IIX, INDEX, J, K, KOFF, L, LM, LMDN INTEGERLMUP, LMX, N, NATX, NATXX, NB, NBM INTEGERJLO,ipass DOUBLE PRECISION HL, RKGM, SX, TPI, check DOUBLE PRECISION ROTV1(3), ROTV2(3), VEC(3) COMPLEX*16 CC ! COMPLEX*16 HH((LOMAX+1)**2*NDF,(LOMAX+1)**2*NDF) COMPLEX*16 HH((2*LOMAX+1)*48,(2*LOMAX+1)*48) COMPLEX*16 SF(NDF), YL(0:(LOMAX+1)**2) !,nv:HSROWS) ! !External Subroutines ! EXTERNAL ROTATE, YLM ! !Intrinsic Functions ! INTRINSIC ATAN, DCMPLX, DCONJG, EXP, SQRT ! ! ** Maybe Experiment ** DOUBLE PRECISION VEC2(3), TMP1, TMP2 ! TPI = 8.0D+0*ATAN(1.0D+0) ! check=2.0D-2 ipass=0 1continue check=check/2.d0 KOFF = NV IA1 = 0 DO 140 N = 1, NAT DO 130 L = 0, LOMAX !IF (LOOR(L,N)) THEN do jlo=1,ilo(l,n) LMDN = L*L + 1 LMUP = (L+1)*(L+1) INDEX = 0 NB = 0 NBM = MULT(N)*(1+LMUP-LMDN) DO 120 IEQ = 1, MULT(N) DO 110 LM = LMDN, LMUP NB = NB + 1 K = KOFF + NB 10CONTINUE INDEX = INDEX + 1 IF (INDEX .GT. NV) GOTO 900 ! WRITE (6,*) 'INDEX,K,N,L,IEQ,LM',INDEX,K,N,L,IEQ,LM KZZ(1,K) = KZZ(1,INDEX) KZZ(2,K) = KZZ(2,INDEX) KZZ(3,K) = KZZ(3,INDEX) XK(K) = XK(INDEX) YK(K) = YK(INDEX) ZK(K) = ZK(INDEX) RKGM = SQRT(XK(K)*XK(K)+YK(K)*YK(K)+ZK(K)*ZK(K)) IF (NBM .NE. 1) THEN DO 20 NATX = 1, MULT(N) NATXX = IA1 + NATX SX = KZZ(1,K)*POS(1,NATXX) + & KZZ(2,K)*POS(2,NATXX) + & KZZ(3,K)*POS(3,NATXX) ! SF(NATX) = EXP(DCMPLX(0.0D+0,TPI*SX)) SF(NATX) = DCMPLX(DCOS(TPI*SX),DSIN(TPI*SX)) 20 CONTINUE IIX = 0 DO 50 NATX = 1, MULT(N) IF (RKGM .LE. 1.0D-5) THEN DO 30 LMX = LMDN, LMUP !YL(LMX-1,K) = (0.0D+0,0.0D+0) YL(LMX-1) = 0.0D0 30 CONTINUE ! YL(0,K) = (1.0D+0,0.0D+0) YL(0) = 1.D0 ELSE VEC(1) = XK(K) VEC(2) = YK(K) VEC(3) = ZK(K) CALL ROTATE(VEC,ROTIJ(1,1,IA1+NATX),ROTV1) CALL ROTATE(ROTV1,ROTLOC(1,1,N),ROTV2) CALL YLM(ROTV2,LOMAX,YL(0)) !,K)) ENDIF DO 40 LMX = LMDN, LMUP IIX = IIX + 1 HH(IIX,NB) = SF(NATX)*YL(LMX-1) !,K) 40 CONTINUE 50 CONTINUE IF (NB .NE. 1) THEN DO 80 J = 1, NB - 1 CC =
Re: [Wien] strange(?) LOPW - Error
Dear Gerhard, This is (at least for me) a well known behavior. We attach to each local orbital a fictitious plane wave with a k-vector K. In addition we require that LOs are orthogonal to each other and this requirement becomes a problem if you have many equivalent atoms (8 Ga in your case) and high l=2 (makes 5 d-LOs per atom) and a small RKmax, because then we run out of K vectors. Maybe we are doing the orthogonality check too stupid and could reduce the requirements when running over equivalent atoms, but I've never tested this in details. In your case, a quick test seems to run through when RKmax = 7.5 or bigger. Alternatively, you may want to split the 8 Ga in 4+4 , but in such a way that more symmetry survives and not just in P1. Peter Am 28.01.2024 um 12:20 schrieb Fecher, Gerhard: Hallo Peter, I observed some unexpected (maybe strange) behaviour of lapw1 (Wien2k Version 23.2 compiled with OneAPI 23.1) I wanted to test if there is a ferro- or antiferromagnetic coupling in MnGa4 1) I started with the regular structure (just spin polarized) I m-3m 1 Mn atom 4 Ga atoms (MnGa4_aexp.struct) this one was running without problems (a small (ferro) magnetic moment of 0.6 seems to exist, I did not check energy and higher precesions), 2) then I made the antiferromagnetic structure P m-3m with 2 Mn atoms and 8 Ga atoms (MnGa4_afm-failes.struct) and named the Mn by 1 and two here the LOPW - Error (at the Ga atom, see below), 3) I splitted the Ga atoms in two groups P 4mm with 2 Mn and 2x4 Ga atoms (MnGa4_afm-2.struct) and named both Mn and Ga toms to distinguish same. this one was running without problems, too (an antiferromagnetic mpoment of +-1.2 muB seems to exist). (indeed, this splitting doesn't make much sense, from symmetry point of view) In all three cases the structure in reduced P1 symmetry has the same lattice parameters and positions (2 Mn and 8 Ga) all calculations were simply initialized with precision 1, the parameters in case.in1 are principally always the same, changing the precesion or other parameters in case 2) did not help. The error appeared independet on the magnetic setting in case.inst (ferro, antiferro, with and without magnetic moment at Ga). (the same LOPW error appears when I replace Ga Z=31 by Ge Z=32 (not with Al or Au)) I wonder why this LOPW error appears in the P m-3m structure but not in the others. (as the P 4mm structure has only 8 symmetry operations it would be faster to run the P m-3m structure (48 symmetry operations)) some detail of the error (note: running lapw1 -dn by hand has the same result as lapw1 -up): the message at the end of case.scf1up (or output1up) is :INFO : LOPW-exhausted for atom3 PASS 1 had to reduce check 0.01 :INFO : LOPW-exhausted for atom3 PASS 2 had to reduce check 0.005000 :INFO : LOPW-exhausted for atom3 PASS 3 had to reduce check 0.002500 :INFO : LOPW-exhausted for atom3 PASS 4 had to reduce check 0.001250 :INFO : LOPW-exhausted for atom3 PASS 5 had to reduce check 0.000625 Ciao I don't expect an answer today, better have a nice weekend, Gerhard ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html -- -- Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna Phone: +43-1-58801-165300 Email: peter.bl...@tuwien.ac.atWIEN2k: http://www.wien2k.at WWW: http://www.imc.tuwien.ac.at - ___ Wien mailing list Wien@zeus.theochem.tuwien.ac.at http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien SEARCH the MAILING-LIST at: http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
