Depending on the options (as keywords) specified in
the Run-Section of the input file, output
files are created in a directory, the name of the directory being
defined in the same section as an argument to the "name"
keyword.
The following files are always created: log, timing, and input.
The following files may be created:
output prodwei dvr iteration natpot cpd vpot vpot_LABEL
This file contains informations such as the type of calculation performed, the potential energy surface used, minimum possible memory sizes for different arrays, the opened data files and any error messages if any. The most important thing in this file is to check the labelling of the parameters of the coordinate system used which should be consistent with the arguments of the potential energy surface used.
This is a copy of the input file used with the
program version progver and the list of options (if
any given) added.
The output will be saved to
file or written to the screen, depending on the
output keyword in the Run-Section. It contains some standard
results like the "Natural weights (see the MCTDH review for
this)" for all the modes. It also prints, after each iteration
cycle, the rms errors of the fit. Finally, total CPU-time, host,
date and time, path of the name-directory and (if specified in
the input) the title of the run are printed.
This file contains various timing statistics of the potential fitting run. For each subroutine, the number of times the routine is called, the total CPU time (user and system) spent in the routine and in all subroutines below it are listed, as well as the CPU time per function call.
Subroutine Calls N CPU/N CPU %CPU iterations 100 324.082 32408.25 97.9
modrefpot 400 63.401 25360.20 76.6
fiterr 101 67.618 6829.42 20.6
getpes 1 616.440 616.44 1.9
coeffs 1001 0.237 237.69 0.7
firstfit 1 69.050 69.05 0.2
loadv 501 0.028 13.89 0.0
diagrs 505 0.003 1.63 0.0
vecs 1 0.700 0.70 0.0
dvr 1 0.000 0.00 0.0
Total (usr,sys,cpu)
33084.3 9.6
33093.9 ( 9h : 11m : 33.9s)
|
Name |
Routine Description |
|
iterations |
The Iteration scheme used for the potential fit |
|
modrefpot |
The "Modified Reference Potential" module, computes at each iteration step the modified reference potential |
|
fiterr |
Computes the fit error |
|
getpes |
Searches for the potential energy surface in the library and writes out the exact potential to the disc |
|
coeffs |
Computes the (long and/or short) coefficients, called from the direct.F and iteration.F subroutines |
|
loadv |
Loads the potential stored in a vpot file into another array |
|
diagrs |
Diagonalizes the (modified) density matrices using the "DSYEV" routine from Lapack library |
|
vecs |
Called from direct.F, this module computes all the NATURAL POTENTIALS |
|
dvr |
Calls the gendvr module |
Depending on the circumstances at runtime a dvr file containing information on the DVR basis, DVR grid, etc. is written. Its structure is described in the documentation on the DVR File of the MCTDH output.
If the keyword iteration is set in
the Run-Section some data concerning the iteration
procedure is output to this file.
The format is:
For each iteration step (in one line): it : iteration step (starts with 0 for initialization step) Wei-rel: relevant and weighted rms-error [eV], Unw-rel: relevant and unweighted rms-error [eV], Unw-all: unweighted rms-error [eV], Max-rel: relevant and maximal absolute error [eV], Max-all: maximal absolut error [eV], Additionally, for each mode: fn : sqrt[trace-sum of natural weights/wsum2 or niw]
If the input file has a Separable-Weight-Section and if the
keyword prodwei is set the one-dimensional separable
weight vectors are output to this file.
The format is:
Number of degrees of freedom
For each mode: Number of primitive grid points
For each mode: Primitive grid points,
Separable weights
This file contains the "Natural potentials (see
the MCTDH review for the definition of Natural potentials)". This
is unformatted and is read by the MCTDH subroutine
rdnatpot to start any MCTDH run. The followings are
written:
HEADER:progver System definition i.e. dvrdef. Date of creation(The following lines show some input informations for documentation purposes)
write(line,'(a,i3,a,20i4)') ' contracted mode:',modc,
+ ', nat.pots:',(potdim(i),i=1,min(ndof,20))
write(unit,iostat=ierr) line(1:80)
write(line,'(a,e12.4,a,18i3)') ' V-cut =', pesmax*27.2114,
+ ' eV, oned:',(onedipot(i),i=1,min(ndof,18))
write(unit,iostat=ierr) line(1:80)
write(line,'(a,e12.4,a,e12.4,a)') ' C-Weights: V-max =',
+ iwvmax*27.2114,' eV, V-min =',iwvmin*27.2114,' eV'
write(unit,iostat=ierr) line(1:80)
write(line,'(a,11(1x,a))') ' Modelabels:',
+ (modelabel(i)(1:slen(modelabel(i))),i=1,min(ndof,11))
write(unit,iostat=ierr) line(1:80)
(Fitting Parameters)
write(unit) modc,(onedipot(f),f=1,ndof)
write(unit) lpconm
(Modes, DOFS are being written now)
write(unit) dentype
write(unit) ndof,nmode
write(unit) (nspfdof(m),m=1,nmode)
do m=1,nmode
write(unit) (spfdof(n,m),n=1,nspfdof(m))
enddo
write(unit) (dofspf(f),f=1,ndof)
write(unit) log2int(lmult),log2int(leb),log2int(lmulpack)
(Write potdim and one-dimensional potentials)
write(unit) (pdim(m),m=1,npmode)
do f=1,ndof
if (onedipot(f) .gt. 0) then
zeig=zort(f)-1
write(unit) (oned(zeig+g),g=1,gdim(f))
endif
enddo
(Write natural potentials for each contracted and uncontracted modes)
! see routines FITOUT and VECOUT
do m=1,npmode
if (m.eq.modc) then ! contracted mode
do i=1,dimaft
do j=1,dimbef
write (unit) (coeff(j,g,i), g=1,dimmodc)
enddo
enddo
else ! other modes
! natural potentials of mode m start at zevec(m)
evecp => evec(zevec(m)) ! this is symbolic
do i=1,potdim(m)
write (unit) (evecp(g,i), g=1,psubdim(m))
enddo
endif
enddo
(In the end the vpot parameters are written to be used in the
“showpot” program for viewing purpose)
This file contains a potential in cpd format. In general it is
generated by mccpd86 or natpot2cpd86. The binary file has the
following structure:
This file contains the exact
potential energy surface (PES) on the product grid (hence it may
become quite large). It is used by potfit to store the exact
potential externally. In addition, with a vpot file one can plot
cuts of the corresponding PES by the help of the showpot program. These files are written in projection runs.
They contain the projected potentials on the reduced product
grids. Their structure is the same as the normal vpot file, so they can be used in other potfit
runs.
HEADER:progver
System definition i.e. dvrdef
Date of creation
(The following lines show some input informations for documentation purposes)
lcount = 4
write(channel,iostat=ierr) lcount
write(line,'(a,i6)') ' candecomp rank:', rank
write(channel,iostat=ierr) line(1:80)
write(line,'(a,ES16.8,a)') ' V-cut =', pesmax*27.2114, ' eV'
write(channel,iostat=ierr) line(1:80)
write(line,'(a,e12.4,a,e12.4,a)') ' C-Weights: V-max =', &
iwvmax*27.2114,' eV, V-min =',iwvmin*27.2114,' eV'
write(channel,iostat=ierr) line(1:80)
write(line,'(a,11(1x,a))') ' Modelabels:', &
(trim(cpd%modelabel(i)),i=1,min(cpd%ndof,11))
write(channel,iostat=ierr) line(1:80)
write(inpot) rank, onedipot
write grddef (see wrgrddef / rdgrddef)
(Write one-dimensional potentials)
do f=1,ndof
if (onedipot(f) .gt. 0) then
write(unit) (oned(zort(f)-1+g),g=1,gdim(f))
endif
enddo
(Write the cpd potential)
do r=1,rank
write(unit) coeff(r)
do m=1, nmode
z = zspp(r,m)
write(unit) (spp(zspp(r,m)-1+g),g=1,subdim(m))
end do
end do
(In the end the vpot parameters are written to be used in the
“showpot” program for viewing purpose and for
consistency with the natpot file.)
The
Vpot File
The
vpot_LABEL Files
System definition
(primitive definition of the system considered
(ndofs etc))
The following data is written by subroutine wrdvrdef:
ndof
(modelabel(f),f=1,ndof)
(gdim(f),f=1,ndof)
mbaspar
(basis(f),f=1,ndof)
(log2int(ldvr(f)),f=1,ndof)
((ipbaspar(i,f),i=1,mbaspar),f=1,ndof)
((rpbaspar(i,f),i=1,mbaspar),f=1,ndof)
((xend(i,f),i=1,2),f=1,ndof)