Program control

The first card of each input should be:


where text is arbitrary. If file 1 is restarted, text must always be the same. The effect of this card is to reset all program counters, etc. If the *** card is omitted, text assumes its default value, which is all blank.

The end of the input is signaled by either an end of file, or a


card. All input following the --- card is ignored.

Alternatively, a job can be stopped at at some place by inserting an EXIT card. This could also be in the middle of a DO loop or an IF block. If in such a case the --- card would be used, an error would result, since the ENDDO or ENDIF cards would not be found.

In contrast to Molpro92 and older versions, the current version of Molpro attempts to recover all information from all permanent files by default. If a restart is unwanted, the NEW option can be used on the FILE directive. The RESTART directive as described below can still be used as in Molpro92, but is usually not needed.


The $r_i$ specify which files are restarted. These files must have been allocated before using FILE cards. There are two possible formats for the $r_i$:

  • a) $0 < r_i < 10$: Restart file $r_i$ and restore all information.
  • b) $$: Restart file $nr$ but truncate before record name.

If all $r_i=0$, then all permanent files are restarted. However, if at least one $r_i$ is not equal to zero, only the specified files are restarted.


  • RESTART; will restart all permanent files allocated with FILE cards (default)
  • RESTART,1; will restart file 1 only
  • RESTART,2; will restart file 2 only
  • RESTART,1,2,3; will restart files 1-3
  • RESTART,2000.1; will restart file 1 and truncate before record 2000.


Insert the contents of the specified file in the input stream. In most implementations the file name given is used directly in a Fortran open statement. If file begins with the character ’/’, then it will be interpreted as an absolute file name. Otherwise, it will be assumed to be a path relative to the directory from which the Molpro has been launched. If, however, the file is not found, an attempt will be made instead to read it relative to the system lib/include directory, where any standard procedures may be found.

If the ECHO option is specified, the included file is echoed to the output in the normal way, but by default its contents are not printed. The included file may itself contain INCLUDE commands up to a maximum nesting depth of 10.

For the name of the file and the characters allowed, similar recommendations hold as for the Molpro input file, see . The include file together with a path to it may however be used.


Sets the limit on dynamic memory to $n$ floating point words. For details, see section memory allocation.

DO loops can be constructed using the DO and ENDDO commands. The general format of the DO command is similar to Fortran:

DO variable=start, end [[,]increment] [[,]unit]

where start, end, increment may be expressions or variables. The default for increment is 1. In contrast to Fortran, these variables can be modified within the loop (to be used with care!). For instance:

DO R=1.0,6.0,DR,ANG
IF (R.EQ.2) DR=0.5
IF (R.EQ.3) DR=1.0

performs the loop for the following values of R: 1.0, 1.2, 1.4, 1.6, 1.8, 2.0, 2.5, 3.0, 4.0, 5.0, 6.0 Ångstrøm. The same could be achieved as follows:

RVEC=[1.0,1.2,1.4,1.6,1.8,2.0,2.5,3.0,4.0,5.0,6.0] ANG

Up to 20 DO loops may be nested. Each DO must end with its own ENDDO.

Jumps into DO loops are possible if the DO variables are known. This can be useful in restarts, since it allows to continue an interrupted calculation without changing the input (all variables are recovered in a restart).

See introductory examples for two examples of using do loops.

IF blocks and IF/ELSEIF blocks can be constructed as in FORTRAN.

IF blocks have the same form as in Fortran:

IF (logical expression) THEN statements ENDIF

If only one statement is needed, the one-line form

IF (logical expression) statement

can be used, except if statement is a procedure name.

ELSE and ELSE IF can be used exactly as in Fortran. IF statements may be arbitrarily nested. Jumps into IF or ELSE IF blocks are allowed. In this case no testing is performed; when an ELSE is reached, control continues after ENDIF.

The logical expression may involve logical comparisons of algebraic expressions or of strings. Examples:

TEXT,An error occurred, calculation stopped
IF($method.eq.'HF') then

In the previous example the dollar and the quotes are optional:


GOTO commands can be used to skip over parts of the input. The general form is


Program control skips to the $|n|$’th occurrence of command (Default: $n=1$). command must be a keyword in the first field of an input line. If n is positive, the search is forward starting from the current position. If n is negative, search starts from the top of the input. The GOTO command is executed at most nrep times. The default for nrep is 1 if $n \lt 0$ and infinity otherwise. We recommend that GOTO commands are never used to construct loops.

Alternatively, one can jump to labels using


Since labels must be unique, the search starts always from the top of the input. It is required that the label ends with a colon.


This is a dummy command, sometimes useful in conjunction with GOTO.

Procedures can be defined at any place of the input or in INCLUDE files as follows:

PROC name 

Alternatively, one can use the form

PROC name[=]{statements}

In the latter case, it is required that the left curly bracket ($\{$) appears on the same line as PROC, but statements can consist of several lines. If in the subsequent input name is found as a command in the first field of a line, it is substituted by the statements. Example:

proc runscf1
if( set,scfsym=symmetry(1)
if( set,scfsym=scfsy
if( set,scfsym=scfsymm
if( set,scfsym=scfsymmetry
if(#scfsym.eq.0) set,scfsym=1
if(orbital.eq.0) then
else if('RHF' \ then
  if(spin(1).eq.0.and.mod(nelec(1),2).ne.0) set,spin(1)=1
end if
if(#spin.eq.0) spin=mod(nelec,2)

Alternatively, this could be written as

proc runscf2={
if( set,scfsym=symmetry(1)
if( set,scfsym=scfsy
if( set,scfsym=scfsymm
if( set,scfsym=scfsymmetry
if(#scfsym.eq.0) set,scfsym=1
if(orbital.eq.0) then
else if('RHF' \ then
  if(spin(1).eq.0.and.mod(nelec(1),2).ne.0) set,spin(1)=1
end if
if(#spin.eq.0) spin=mod(nelec,2)

Procedures may be nested up to a depth of 10. In the following example runscf is a procedure:

proc runmp2
if(spin.eq.0) then
end if

Note: Procedure names are substituted only if found in the first field of an input line. Therefore, they must not be used on one-line IF statements; please use IF / ENDIF structures instead.

If as first statement of a procedure ECHO is specified, the substituted commands of the present and lower level procedures will be printed. If ECHO is specified in the main input file, all subsequent procedures are printed.

Certain important input data can be passed to the program using variables. For instance, occupancy patterns, symmetries, number of electrons, and multiplicity can be defined in this way (see section special variables for more details). This allows a quite general use of procedures. Procedures can also be used in geometry optimizations or extrapolations, for example:

runmrci={if(.not.scfdone) hf;multi;mrci}


Various procedures for standard calculations can be used with the run command, see Simplified input. They allow to write simple inputs in the most compact way.

Further predefined procedures are available in lib/include/procedures:

runscf, rundft, runmp2, rundf-hf, rundt-mp2, runldf-hf, runldf-hf, runpno-lmp2, runmp3, runmp4, runmp4sdq,  runccsd, runccsdt, runbccd, runbccdt, runqcisd runqcisdt, runuccsd, runuccsdt, runcas, runmrpt, runcaspt2, runcaspt3, runmrci, runacpf, optscf, optdft, optmp2, optcas, optmrci, optccsd, optccsdt, freqscf, freqdft, freqmp2, freqccsd, freqccsdt, freqcas, freqmrci, rundf-lmp2, optdf-lmp2, freqdf-lmp2

A line

include procedures

is necessary to include these procedures. Some procedures use variables SYMMETRY, SPIN, STATE, NELEC to define state symmetries, spins, and charges, respectively. For example, the procedure runmrci can be used for a calculation of a vertical ionization potential of H$_2$O as follows:

***,h2o IP
include procedures
r=1 ang           !set bond distance
theta=104 degree  !set bond angle

basis=vtz         !define basis set

geometry          !geometry input block
o                 !z-matrix
endg              !end of geometry input
runmrci           !compute mrci energy of water using defaults
eh2o=energy       !save mrci energy in variable eh2o

set,nelec=9       !set number of electrons to 9
set,symmetry=2    !set wavefunction symmetry to 2
runmrci           !compute mrci energy of h2o+ (2b2 state)

ipci=(energy-eh2o)*toev  !compute mrci ionization potential in ev

For further examples see oh_runccsdt.inp, oh_runmrci1.inp, oh_runmrci2.inp, oh_runmrci3.inp, oh_runmrci4.inp.

Note: At present, all variables are global, i.e., variables are commonly known to all procedures and all variables defined in procedures will be subsequently known outside the procedures as well. The reason is that procedures are included into the internal input deck at the beginning of the job and not at execution time; for the same reason, variable substitution of procedure names is not possible, e.g. one cannot use constructs like

$method       !this does not work!


will just print xxxxxx in the output. If the text contains variables which are preceded by a dollar ($), these are replaced by their actual values, e.g.

text,Results for R=$r

will print

Results for R=2.1


This command checks and prints the status of the specified program steps. commands may be a list of commands for wavefunction calculations previously executed in the current job. If no command or LAST is specified, the status of the last step is checked. If ALL is given, all program steps are checked.

If CRASH or STOP is given, the program will crash or stop, respectively, if the status was not o.k. (STOP is default). If IGNORE is given, any bad status is ignored. If CLEAR is specified, all status information for the checked program steps is erased, so there will be no crash at subsequent status checks.


  • STATUS,HF,CRASH; will check the status of the last HF-SCF step and crash if it was not o.k. (i.e. no convergence). CRASH is useful to avoid that the next program in a chain is executed.
  • STATUS,MULTI,CI,STOP; will check the status of the most previous MULTI and CI steps and stop if something did not converge.
  • STATUS,RHF,CLEAR; will clear the status flag for last RHF. No action even if RHF did not converge.

Note that the status variables are not recovered in a restart.

By default, the program automatically does the following checks:

1.) If an orbital optimization did not converge, and the resulting orbitals are used in a subsequent correlation calculation, an error will result. This error exit can be avoided using the IGNORE_ERROR option on the ORBITAL directive.

2.) If a CCSD|QCI|BCC|LMPn calculation did not converge, further program steps which depend on the solution (e.g, Triples, CPHF, EOM) will not be done and an error will result. This can be avoided using the NOCHECK option on the command line.

3.) In geometry optimizations or frequency calculations no convergence will lead to immediate error exits.


A number of global thresholds can be set using the GTHRESH command outside the individual programs (the first letter G is optional, but should be used to avoid confusion with program specific THRESH cards). The syntax is


key can be one of the following.

  • ZERO Numerical zero (default 1.d-12)
  • ONEINT Threshold for one-electron integrals (default 1.d-12, but not used at present)
  • TWOINT Threshold for the neglect of two-electron integrals (default 1.d-12)
  • PREFAC Threshold for test of prefactor in TWOINT (default 1.d-14)
  • LOCALI Threshold for orbital localization (default 1.d-8)
  • EORDER Threshold for reordering of orbital after localization (default 1.d-4)
  • ENERGY Convergence threshold for energy (default 1.d-6)
  • GRADIENT Convergence threshold for orbital gradient in MCSCF (default 1.d-2)
  • STEP Convergence threshold for step length in MCSCF orbital optimization (default 1.d-3)
  • ORBITAL Convergence threshold for orbital optimization in the SCF program (default 1.d-5).
  • CIVEC Convergence threshold for CI coefficients in MCSCF and reference vector in CI (default 1.-d.5)
  • COEFF Convergence threshold for coefficients in CI and CCSD (default 1.d-4)
  • PRINTCI Threshold for printing CI coefficients (default 0.05)
  • PUNCHCI Threshold for punching CI coefficients (default 99 - no punch)
  • SYMTOL Threshold for finding symmetry equivalent atoms (default 1.d-6)
  • GRADTOL Threshold for symmetry in gradient (default 1.d-6).
  • THROVL Threshold for smallest allowed eigenvalue of the overlap matrix (default 1.d-8)
  • THRORTH Threshold for orthonormality check (default 1.d-8)
  • THRPRINT Threshold for printing orbitals (thrprint=-1 : column-wise; thrprint=0 : row-wise, as in Molpro2015 and earlier versions ; thrprint $>0$: print only coefficients that are larger than the threshold together with labels (default: thrprint=0.25)


Global print options can be set using the GPRINT command outside the individual programs (the first letter G is optional, but should be used to avoid confusion with program specific PRINT cards). The syntax is


Normally, value can be omitted, but values $\gt 0$ may be used for debugging purposes, giving more information in some cases. The default is no print for all options, except for DISTANCE, ANGLES (default=0), and VARIABLE. NOGPRINT,key is equivalent to PRINT,key=-1. key can be one of the following:

  • BASIS Print basis information
  • DISTANCE Print bond distances (default)
  • ANGLES Print bond angle information (default). If $\gt$ 0, dihedral angles are also printed.
  • ORBITAL Print orbitals in SCF and MCSCF
  • ORBEN Print orbital energies in SCF
  • CIVECTOR Print CI vector in MCSCF
  • PAIRS Print pair list in CI, CCSD
  • CS Print information for singles in CI, CCSD
  • CP Print information for pairs in CI, CCSD
  • REF Print reference CSFs and their coefficients in CI
  • PSPACE Print p-space configurations
  • MICRO Print micro-iterations in MCSCF and CI
  • CPU Print detailed CPU information
  • IO Print detailed I/O information
  • VARIABLE Print variables each time they are set or changed (default).

The operators for which expectation values are requested, are specified by keywords on the global GEXPEC directive. By default, only dipole moments are computed. The first letter G is optional, but should be used to avoid confusion with program specific EXPEC cards, which have the same form as GEXPEC. For all operators specified on the GEXPEC card, expectation values are computed in all subsequent programs that generate the first-order density matix. This is always the case for variational wavefunctions, i.e., HF, DFT, MCSCF, MRCI. For non-variational wavefunctions such as MP2, MP3, QCISD, QCISD(T), CCSD, or CCSD(T) the density matix is not computed by default, since this requires considerable additional effort (solving z-vector equations). The GEXPEC directive does not affect such programs. In some cases [currently for MP2, MP3, QCISD, QCISD(T), and CCSD] the EXPEC directive that is specific to those programs can be used to request the property calculation.

For a number of operators it is possible to use generic operator names, e.g., DM for dipole moments, which means that all three components DMX, DMY, and DMZ are computed. Alternatively, individual components may be requested.

The general format is as follows:



  • opname operator name (string), either generic or component.
  • icen z-matrix row number or z-matrix symbol used to determine the origin (x,y,z must not be specified).

If icen$=0$ or blank, the origin must be specified in x,y,z

Several GEXPEC cards may follow each other, or several operators may be specified on one card.


GEXPEC,QM computes quadrupole moments with origin at (0,0,0),

GEXPEC,QM1 computes quadrupole moments with origin at centre 1.

GEXPEC,QM,O1 computes quadrupole moments with origin at atom O1.

GEXPEC,QM,,1,2,3 computes quadrupole moments with origin at (1,2,3).

The following table summarizes all available operators:

One-electron operators and their components
Generic name Parity Components Description
OV 1 Overlap
EKIN 1 Kinetic energy
POT 1 potential energy
DELTA 1 delta function
DEL4 1 $\Delta^4$
DARW 1 one-electron Darwin term, i.e., DELTA with appropriate factors summed over atoms.
MASSV 1 mass-velocity term, i.e., DEL4 with appropriate factor.
REL 1 total Cowan-Griffin Relativistic correction, i.e., DARW+MASSV.
DM 1 DMX, DMY, DMZ dipole moments
SM 1 XX, YY, ZZ, XY, XZ, YZ second moments
TM 1 XXX, XXY, XXZ, XYY, XYZ, XZZ, YYY, YYZ, YZZ, ZZZ third moments
MLTPn 1 all unique Cartesian products of order $n$ multipole moments
QM 1 QMXX, QMYY, QMZZ, QMXY, QMXZ, QMYZ, QMRR=XX + YY + ZZ, QMXX=(3 XX - RR)/2, QMXY=3 XY / 2 etc. quadrupole moments and $R^2$
EF 1 EFX, EFY, EFZ electric field
FG 1 FGXX, FGYY, FGZZ, FGXY, FGXZ, FGYZ electric field gradients
DMS 1 DMSXX, DMSYX, DMSZX, DMSXY, DMSYY, DMSZY, DMSXZ, DMSYZ, DMSZZ diamagnetic shielding tensor
LOP -1 LX, LY, LZ Angular momentum operators $\hat L_x$, $\hat L_y$, $\hat L_z$
LOP2 1 LXLX, LYLY, LZLZ, LXLY, LXLZ, LYLZ one electron parts of products of angular momentum operators. The symmetric combinations $\frac{1}{2} (\hat L_x \hat L_y+\hat L_y \hat L_x)$ etc. are computed.
VELO -1 D/DX, D/DY, D/DZ velocity
LS -1 LSX, LSY, LSZ spin-orbit operators
ECPLS -1 ECPLSX, ECPLSY, ECPLSZ ECP spin-orbit operators

Expectation values are only nonzero for symmetric operators (parity=1). Other operators can be used to compute transition quantities (spin-orbit operators need a special treatment).

The following job computes dipole and quadrupole moments for H$_2$O.

***,h2o properties
geometry={o;h1,o,r;h2,o,r,h1,theta}   !Z-matrix geometry input
r=1 ang                               !bond length
theta=104                             !bond angle
gexpec,dm,sm,qm                       !compute dipole and quarupole moments
$methods=[hf,multi,ci]                !do hf, casscf, mrci
do i=1,#methods                       !loop over methods
$methods(i)                           !run energy calculation
dip(i)=dmz                            !save dipole moment in variable dip
quadxx(i)=qmxx                        !save quadrupole momemts
smxx(i)=xx                            !save second momemts
table,methods,dip,smxx,smyy,smzz        !print table of first and second moments
table,methods,e,quadxx,quadyy,quadzz    !print table of quadrupole moments

This Job produces the following tables

 METHODS     DIP           SMXX          SMYY          SMZZ
 HF        0.82747571   -5.30079792   -3.01408114   -4.20611391
 MULTI     0.76285513   -5.29145148   -3.11711397   -4.25941000
 CI        0.76868508   -5.32191822   -3.15540500   -4.28542917

 METHODS       E             QUADXX       QUADYY        QUADZZ
 HF        -76.02145798   -1.69070039   1.73937477   -0.04867438
 MULTI     -76.07843443   -1.60318949   1.65831677   -0.05512728
 CI        -76.23369821   -1.60150114   1.64826869   -0.04676756
geometry={ar1;ar2,ar1,r}   !geometry definition
r=2.5 ang                  !bond distance
{hf;                       !non-relativisitic scf calculation
expec,rel,darwin,massv}    !compute relativistic correction using Cowan-Griffin operator
e_nrel=energy              !save non-relativistic energy in variable enrel
show,massv,darwin,erel     !show individual contribution and their sum

dkroll=1                   !use douglas-kroll one-electron integrals
hf;                        !relativistic scf calculation
e_dk=energy                !save relativistic scf energy in variable e_dk.
show,massv,darwin,erel     !show mass-velocity and darwin contributions and their sum
show,e_dk-e_nrel           !show relativistic correction using Douglas-Kroll

This jobs shows at the end the following variables:

 MASSV / AU             =       -14.84964285
 DARWIN / AU            =        11.25455679
 EREL / AU              =        -3.59508606

This command allows to modify the default behaviour of the XML output. The general format is as follows:


where key can be one of the following:

  • DUMPORB Dump occupied orbitals to XML file (default 0, unless the –xml-orbdump command line option is given).
  • SKIPVIRT Omit virtual orbitals in the XML orbital dumps (default 1).
  • PRINT Print record information to output file, showing from which records the orbitals etc are read (default 0).