Murnaghan fitting program
PROGRAM MURN
C
C modified from murn2.f : R.S 19.04.91
C modified from murn1.f : gps 6.12.90
C least-squares fit of E vs V by Murnaghan equation of state
C
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION EDATA(50),VDATA(50),X(40),AUX(40)
EXTERNAL FUN
C CONVERSION FACTOR FROM RYDBERG TO EV:
data CONVYY /13.605826/
COMMON /C/ EDATA,VDATA, VOLMIN,VOLMAX,B0MIN,B0MAX,B0PMIN,B0PMAX,U
&LA,NNN
C
IN=5
IOUT=6
write(iout,*) 'least-squares fit of etot(v) by Murnaghan eq.'
write(iout,*)
c WRITE(IOUT,1002)
c 1002 FORMAT(' LEAST-SQUARES FIT OF ETOT(V) BY MURNAGHAN EQUATION'/1X)
C
C READING THE INPUT DATA:
C
C UNIT OF LENGTH, IN ANGSTROEM, WHICH WILL BE USED THROUGHOUT
C THIS PROGRAM:
ULA=0.52917715
WRITE(IOUT,1010)ULA
1010 FORMAT(12X,'UNIT OF LENGTH: ULA =',F15.6,2X,'ANGSTROEM')
C
C UNITS OF ENERGY USED TO INPUT THE DATA: 1 MEANS RYDBERG, 2 IS EV., 3
C = HARTREE
C UNITS OF
READ(IN,*)IUNIT
WRITE(IOUT,1060)IUNIT
1060 FORMAT(12X,'UNIT OF ENERGY: IUNIT =',I4,':')
IF(IUNIT .EQ. 1) then
CONV_inp = CONVYY
WRITE(IOUT,1070)
else IF(IUNIT .EQ. 2) then
CONV_inp = 1.0
WRITE(IOUT,1080)
else IF(IUNIT .EQ. 3) then
CONV_inp = 2 * CONVYY
WRITE(IOUT,1081)
else
WRITE(IOUT,1090)
STOP
ENDIF
1070 FORMAT(40X,' (ENERGIES INPUT IN RYDBERGS)'/1X)
1080 FORMAT(40X,' (ENERGIES INPUT IN ELECTRONVOLTS)'/1X)
1081 FORMAT(40X,' (ENERGIES INPUT IN HARTREES)'/1X)
1090 FORMAT(' ONLY IUNIT = 1 AND 2 ARE RECOGNIZED; STOP.')
C
C CONVERSION FACTOR FROM A**3 TO THE VOLUME OF UNIT CELL
C (E.G. 0.25, IF THE STRUCTURE IS SUCH THAT V = A**3/4 ):
READ(IN,*)CONVAV
WRITE(IOUT,1020)CONVAV
1020 FORMAT(' IN THE STRUCTURE CONSIDERED, THE UNIT CELL VOLUME = A**3
& *',F15.6/1X)
c read in alat boundaries and number of points for that fitted Etot sho
c uld be computed
c read in
read(in,*) alat_min, alat_max, nr_alat
write(iout,
& '("min and max alat(bohr) to compute Etot for at",i4," points ="
&,f20.7, " and", f20.7)')
& nr_alat, alat_min, alat_max
if (nr_alat .lt. 2 .or. alat_max .le. alat_min) stop 'something w
&rong with this range'
C NUMBER OF DATA POINTS A, E(A):
READ(IN,*) NNN
WRITE(IOUT,1030)NNN
1030 FORMAT(9X,'NUMBER OF DATA POINTS: N =',I5)
IF(NNN .GT. 50) THEN
WRITE(IOUT,*)' N TOO LARGE, NMAX=50'
STOP
ENDIF
C
WRITE(IOUT,1120)
WRITE(IOUT,1040)
1040 FORMAT(1X/5X,'I',5X,'ALATT',9X,'VOLUME',8X,
1 'ENERGY (EV)',12X,'ENERGY (RY)'/1X)
C
DO 10 I=1,NNN
READ(IN,*)ALATT,ENINP
ALATT=ALATT
VDATA(I)=ALATT**3*CONVAV
EDATA(I)=ENINP*CONV_inp
WRITE(IOUT,1110)I,ALATT,VDATA(I),EDATA(I),ENINP
1110 FORMAT(1X,I5,2F15.6,2F19.10)
10 CONTINUE
WRITE(IOUT,1120)
1120 FORMAT(1X,79(1H-)/1X)
C
C STARTING VALUES FOR ITERATION:
C
WRITE(IOUT,1130)
1130 FORMAT(6X,'ALATT0',3X, 'VOL0 (ULA**3)',2X,'B0 (MBAR)',5X,'B0PRIME
&',6X,'E0 (EV)',6X,'E0 (RY)'/1X)
C
C MIMIMUM IN EDATA(I):
E0MIN=1.D6
VOLMIN=1.D6
DO 100 I=1,NNN
IF(EDATA(I) .GE. E0MIN) GO TO 100
E0MIN=EDATA(I)
VOLMIN=VDATA(I)
100 CONTINUE
C
VOL0=VOLMIN
E0EV=E0MIN
E0RY=E0*CONVYY
ALATT0=(VOL0/CONVAV)**(1.D0/3.D0)
C
C FOR OTHER VARIABLES, WE CHOOSE:
c
B0=1.D0
B0PRIM=4.D0
C
WRITE(IOUT,1140)ALATT0,VOL0,B0,B0PRIM,E0EV,E0RY
1140 FORMAT(5X,F9.4,3F13.5,2F13.5)
WRITE(IOUT,1150)
1150 FORMAT(64X,'(=INITIAL GUESS)'/1X)
C
ALMIN=0.5D0*ALATT0
ALMAX=1.5D0*ALATT0
B0MIN=0.01D0
B0MAX=10.D0
B0PMIN=0.01D0
B0PMAX=10.D0
C
VOLMIN=ALMIN**3.D0*CONVAV
VOLMAX=ALMAX**3.D0*CONVAV
WRITE(IOUT,1210)ALMIN,VOLMIN,B0MIN,B0PMIN,ALMAX,VOLMAX,B0MAX,B0PM
&AX
1210 FORMAT(' MIN:',F9.4,3F13.5/' MAX:',F9.4,3F13.5)
C
C A UNIFORM SHIFT OF ALL THE ENERGIES
C (WILL NOT INFLUENCE THE B0, B0PRIME, VOL0, ONLY SHIFTS E0):
C
SHIFT=-E0EV-1.D0
DO 20 I=1,NNN
20 EDATA(I)=EDATA(I)+SHIFT
C
C MURNAGHAN LEAST-SQUARES FIT:
WRITE(IOUT,1120)
WRITE(IOUT,1280)
1280 FORMAT(' FIT BY MURNAGHAN EQUATION EQUATION OF STATE')
WRITE(IOUT,1120)
WRITE(IOUT,1190)
1190 FORMAT(1X,'ITER',1X,'ALATT0',3X, 'VOL0 (ULA**3)',2X,
1 'B0 (MBAR)',5X,'B0PRIME',6X,'E0 (EV)',4X,'SUM OF SQUARES'/
2 75X,'IERR'/1X)
C
X(1)=E0EV+SHIFT
X(2)=B0
X(3)=B0PRIM
X(4)=VOL0
NVAR=4
LIM=1!20
C
DO 70 I = 1, 75
CALL DFMND(FUN,X,FFF,NVAR,LIM,AUX,IERR)
C
C THE RESULT OF 20 ITERATION :
WRITE(IOUT,1250) LIM*I,(X(4)/CONVAV)**(1.D0/3.D0),X(4),X(2),X(3),
1 X(1)-SHIFT,FFF,IERR
1250 FORMAT(1X,I4,F9.4,5F13.5/64X,I15)
C
IF(IERR .EQ. 0) GO TO 80
C
70 CONTINUE
80 CONTINUE
C THE RESULT OF THE LAST ITERATION:
WRITE(IOUT,1250)LIM*I,(X(4)/CONVAV)**(1.D0/3.D0),X(4),X(2),X(3),
1 X(1)-SHIFT,FFF,IERR
C
WRITE(IOUT,1120)
WRITE(IOUT,1130)
VOL0=X(4)
ALATT0=(VOL0/CONVAV)**(1.D0/3.D0)
B0=X(2)
B0PRIM=X(3)
E0EV=X(1)-SHIFT
E0RY=E0EV/CONVYY
write(iout,'("alat=",f12.7, " b0=", f10.4, " b0p=", f10.3," E0=",
&f18.6)')
& alatt0, b0, b0prim, E0EV / (2 * CONVYY)
WRITE(IOUT,1140)ALATT0,VOL0,B0,B0PRIM,E0EV,E0RY
alatt0=ula*alatt0
write(iout,1140)alatt0
WRITE(IOUT,1120)
c
c print out fitted energy values for nr_alat lattice
c constants around the equilibrium value given by alat_min and alat
c _max
c cons
c
write(iout,500)
500 format(8x,'alat(A)',10x,'energy(Ryd)',12x,'alat(bohr)', 10x,'ener
&gy(H)')
write(iout,1120)
do i=1,nr_alat
c vol=0.6d0*vol0 + (i-1)*0.05*vol0
c alatt=(vol/convav)**(1./3.)
c alatt=alatt*ula
alatt=alat_min + (i-1)*(alat_max-alat_min)/(nr_alat-1)
vol = alatt**3 * convav
alatt=alatt*ula
call murng1(ula,vol,vol0,b0,b0prim,e0ev,etot)
etotry=etot/convyy
write(iout,502) alatt, etotry, alatt/ula, etotry/2
enddo
502 format(2(f15.5,f20.7))
c
STOP
END
C - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
DOUBLE PRECISION FUNCTION FUN(X)
C FUNCTION TO BE MINIMIZED IN THE LEAST-SQUARES FIT (BY SUBROUTINE DFMN
C D)
C FUNCTION
C CALCULATES THE SUM OF THE A B S O L U T E DEVIATIONS
C (E(THEOR)-E(EXP))**2
C DIVIDED BY THE NUMBER OF EXP. POINTS,
C ASSUMING FOR EQUATION OF STATE THE MURNAGHAN EXPRESSION.
C
C MEANING OF VARIABLES:
C X(1) .... E0
C X(2) .... B0
C X(3) .... B0PRIM
C X(4) .... VOL0
C
C
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
DIMENSION EDATA(50),VDATA(50),X(40)
COMMON /C/ EDATA,VDATA, VOLMIN,VOLMAX,B0MIN,B0MAX,
1 B0PMIN,B0PMAX,ULA,NNN
C
C * * * * * *
E0=X(1)
B0=X(2)
B0PRIM=X(3)
VOL0=X(4)
C
SUM=0.D0
C THE SUM OF SQUARES:
DO 10 I=1,NNN
VOLACT=VDATA(I)
CALL MURNG1(ULA,VOLACT,VOL0,B0,B0PRIM,E0,ETOT)
SUM=SUM+(ETOT-EDATA(I))**2
10 CONTINUE
FUN=SUM/DFLOAT(NNN)
RETURN
END
C -----------------------------------------------------------------
SUBROUTINE MURNG1(ULA,VOL,VOL0,B0,B0PRIM,E0,ETOT)
C EVALUATION OF THE MURNAGHAN EXPRESSION FOR ENERGY AS A FUNCTION
C OF VOLUME.
C
C INPUT DATA:
C
C ULA ..... UNIT OF LENGTH, IN ANGSTROEMS, USED HERE.
C VOL ..... VOLUME, IN THE ABOVE UNITS OF LENGTH CUBED.
C VOL0 .... VOLUME AT THE ENERGY MINIMUM, IN THE SAME UNITS.
C B0 ...... BULK MODULUS, IN UNITS MEGABAR.
C B0PRIM .. PRESSURE DERIVATIVE OF THE BULK MODULUS.
C SHOULD BE CLOSE NEITHER TO 0 NOR TO 1.
C E0 ...... AN ADDITIVE CONSTANT (IN ELECTRONVOLTS), ADDED
C TO THE ENERGY EXPRESSION.
C (SEE, PR B28, p 5484: Fu and Ho)
C
C OUTPUT DATA:
C
C ETOT .... THE ENERGY, INCLUDING THE E0 CONSTANT, IN ELECTRONVOLTS
C
C IF B0 DIFFERS FROM 0. OR FROM 1. BY LESS THAN 1.d-6, THEN
C ETOT IS SET AT +111111.111111 ELECTRONVOLTS.
C
C * * * * * * * *
C
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
C
C CONVERSION FACTOR FROM ERG TO ELECTRONVOLT:
PARAMETER( CONVXX = 1.60209D0 )
C 1 ELECTRONVOLT = 1.60209 D-12 ERG
C
C
IF(DABS(B0PRIM)-1.D0 .LT. 1.d-6 .OR.
1 DABS(B0PRIM) .LT. 1.d-6) THEN
ETOT=+111111.111111D0
RETURN
END IF
C
IF(VOL .LT. 0.D0 .OR. VOL0 .LT. 0.D0) THEN
ETOT=+111111.111111D0
RETURN
END IF
C
ETOT = E0 - B0*VOL0/B0PRIM *
1 (((VOL/VOL0)**(1.D0-B0PRIM)-1.D0)/(1.D0-B0PRIM)-VOL/VOL0+1.D0)
2 *ULA**3/CONVXX
C
RETURN
END
C ---------------------------------------------------------------------
C --
C --------
SUBROUTINE DFMND(F,X,Y,N,LIM,AUX,IER)
C
C ******************************************************************
C * MINIMIZATION OF A FUNCTION OF SEVERAL VARIABLES *
C * USING POWELL'S ALGORITHM WITHOUT DERIVATIVES *
C ******************************************************************
C
DOUBLE PRECISION F,X,Y,AUX,EPS,ETA,TOL,
1 DA,DB,DC,DM,DQ,FA,FB,FC,FL,FM,FS,HD,HQ,HX
DIMENSION X(1),AUX(1)
C
C SUBROUTINES REQUIRED: THE EXTERNAL FUNCTION F.
C
C INPUT DATA:
C F .... THE FUNCTION OF N VARIABLES X(1)...X(N) TO BE MINIMIZED
C X .... X(1) ... X(N) = STARTING GUESS FOR THE VARIABLES;
C BEWARE: X HAS TO BE DIMENSIONED IN THE MAIN PROGRAM
C TO CONSIDERABLY MORE THAN N.
C N .... NUMBER OF VARIABLES; THE DIMENSION OF X AND AUX IN THE
C CALLING PROGRAM MUST BE, HOWEVER, MUCH HIGHER:
C - PERHAPS 10 TIMES?
C LIM .. MAXIMUM NUMBER OF ITERATIONS
C AUX .. AUXILIARY ARRAY OF THE SAME DIMENSION AS X.
C OUTPUT DATA:
C X .... X(1) ... X(N) THE RESULTING MINIMUM
C Y .... VALUE OF THE FUNCTION AT THE MINIMUM
C IER .. SOME ERROR INDICATION
C IERR=0 MEANS 'CONVERGENCE ACHIEVED'.
C
C * * * * * * * *
C
ISW =IER
IER =0
IF (N) 1,1,2
1 IER =1000
GOTO 109
2 IF (LIM) 3,3,4
3 IER =2000
GOTO 109
C
C SET INITIAL VALUES AND SAVE INITIAL ARGUMENT
C
4 N1 =N+1
N2 =N+N
N3 =N+N2
N4 =N+N3
N5 =N*N+N4
EPS =1.D-15
ETA =N*EPS
DO 5 K=1,N
AUX(K)=X(K)
J =N3+K
5 AUX(J)=1.D0
FS =F(AUX)
FB =FS
I =1
IC =1
IT =0
M =N4
MF =0
IS =1
C
C START ITERATION CYCLE
C
6 IT =IT+1
FL =FS
DO 7 K=N1,N2
7 AUX(K)=0.D0
ID =I
I =IC
IW =0
C
C START MINIMIZATION ALONG NEXT DIRECTION
C
8 NS =0
IP =0
DB =0.D0
IF (IW) 10,9,10
9 HX =AUX(I)
10 IF (IT-1) 11,11,14
11 IF (IS-1) 14,12,14
12 DM =.1D0
IF (DABS(HX)-1.D0) 38,38,13
13 DM =-DM*HX
GOTO 38
14 IF (IS-2) 18,15,18
15 IF (IT-1) 17,16,17
16 DM =HQ
GOTO 38
17 DM =DQ
GOTO 38
C
C INTERPOLATION USING ESTIMATE OF SECOND DERIVATIVE
C
18 IF (IW-1) 20,19,20
19 J =N2+I
GOTO 21
20 J =N3+I
21 HD =AUX(J)
DC =1.D-2
IF (IT-2) 23,23,22
22 DC =HQ
23 DM =DC
MK =1
GOTO 51
24 DM =DC*HD
IF (DM) 26,25,26
25 DM =1.D0
26 DM =.5D0*DC-(FM-FB)/DM
MK =2
IF (FM-FB) 27,29,29
27 FC =FB
FB =FM
DB =DC
IF (DM-DB) 28,67,28
28 DC =0.D0
GOTO 51
29 IF (DM-DB) 31,30,31
30 DA =DC
FA =FM
GOTO 37
31 FC =FM
GOTO 51
C
C ANALYSE INTERPOLATED FUNCTION VALUE
C
32 IF (FM-FB) 34,33,33
33 DA =DM
FA =FM
GOTO 35
34 DA =DB
FA =FB
DB =DM
FB =FM
35 IF ((DC-DA)/(DB-DA)) 36,36,50
36 IF (DB) 67,37,67
37 NS =1
DM =-DC
C
C LINEAR SEARCH FOR SMALLER FUNCTION VALUES
C ALONG CURRENT DIRECTION
C
38 IF (NS-15) 43,43,39
39 IF (FS-FM) 41,40,41
40 MF =N+2
DB =0.D0
GOTO 67
41 IF (DABS(DM)-1.D6) 43,43,42
42 IER =100
GOTO 67
43 NS =NS+1
MK =3
GOTO 51
44 IF (FM-FB) 45,46,47
45 DA =DB
FA =FB
DB =DM
FB =FM
DM =DM+DM
GOTO 38
46 IF (FS-FB) 47,45,47
47 IF (NS-1) 48,48,49
48 DA =DM
FA =FM
DM =-DM
GOTO 38
49 DC =DM
FC =FM
C
C REFINE MINIMUM USING QUADRATIC INTERPOLATION
C
50 HD =(FC-FB)/(DC-DB)+(FA-FB)/(DB-DA)
DM =.5D0*(DA+DC)+(FA-FC)/(HD+HD)
IP =IP+1
MK =4
C
C STEP ARGUMENT VECTOR AND CALCULATE FUNCTION VALUE
C
51 IF (IW-1) 54,52,54
52 DO 53 K=1,N
L =M+K
53 AUX(K)=X(K)+DM*AUX(L)
GOTO 55
54 AUX(I)=HX+DM
55 FM =F(AUX)
GOTO (24,32,44,56),MK
C
C ANALYSE INTERPOLATED FUNCTION VALUE
C
56 IF (FM-FB) 61,61,57
57 IF (IP-3) 58,62,62
58 IF ((DC-DB)/(DM-DB)) 60,60,59
59 DC =DM
FC =FM
GOTO 50
60 DA =DM
FA =FM
GOTO 50
61 DB =DM
FB =FM
C
C CALCULATE NEW ESTIMATE OF SECOND DERIVATIVE
C ALONG THE CURRENT DIRECTION
C
62 HD =(HD+HD)/(DC-DA)
IF (IW-1) 64,63,64
63 J =N2+I
GOTO 65
64 J =N3+I
65 AUX(J)=HD
IF (FB-FS) 67,66,67
66 DB =0.D0
C
C SAVE ARGUMENT VECTOR WITH SMALLEST FUNCTION VALUE FOUND
C
67 IF (IW-1) 70,68,70
68 DO 69 K=1,N
L =M+K
J =N+K
HD =DB*AUX(L)
AUX(J)=AUX(J)+HD
HD =X(K)+HD
AUX(K)=HD
69 X(K) =HD
GOTO 71
70 J =N+I
AUX(J)=AUX(J)+DB
HD =HX+DB
AUX(I)=HD
X(I) =HD
71 IF (IER-100) 72,108,72
C
C DETERMINE DIRECTION FOR NEXT LINEAR SEARCH
C
72 FS =FB
IF (I-N) 74,73,73
73 I =0
74 I =I+1
IF (IS) 75,75,80
75 IF (DB) 77,76,77
76 IF (I-IC) 8,77,8
77 IC =I
IS =1
IF (IT-N) 79,79,78
78 IW =1
79 I =ID
GOTO 8
80 M =M+N
IF (M-N5) 82,81,81
81 M =N4
82 IF (IS-1) 83,83,94
83 IF (I-1) 84,84,85
84 IW =1
85 IF (I-ID) 8,86,8
86 HQ =0.D0
DO 87 K=N1,N2
87 HQ =HQ+AUX(K)*AUX(K)
IF (HQ) 90,88,90
88 IF (MF-N1) 108,108,89
89 IER =200
GOTO 108
90 DQ =DSQRT(HQ)
HQ =DQ
IF (HQ-1.D0) 92,92,91
91 HQ =1.D0
92 DO 93 K=N1,N2
L =M+K-N
93 AUX(L)=AUX(K)/DQ
IS =2
GOTO 8
C
C END OF ITERATION CYCLE
C TEST FOR TERMINATION OF MINIMIZATION
C
94 IS =0
TOL =EPS
IF (DABS(FS)-1.D0) 96,96,95
95 TOL =EPS*DABS(FS)
96 IF (FL-FS-TOL) 100,100,97
97 IF (IT-LIM) 99,99,98
98 IER =10
GOTO 108
99 MF =0
GOTO 6
100 IF (MF-N1) 102,102,101
101 IER =200
GOTO 108
102 MF =MF+1
DQ =0.D0
DO 105 K=1,N
J =N+K
IF (DABS(AUX(K))-1.D0) 103,103,104
103 DQ =DQ+DABS(AUX(J))
GOTO 105
104 DQ =DQ+DABS(AUX(J)/AUX(K))
105 CONTINUE
IF (DQ-ETA) 108,108,106
106 IF (MF-N) 6,6,107
107 IER =1
108 Y =FB
IF (IER) 111,111,109
109 IF (ISW+12345) 110,111,110
C 110 CALL WIER(IER,20212)
110 CONTINUE
111 RETURN
END
