A variety of output files are created in the directory specified by the file_autopath keyword in the input file (see Optional Input Entries). All files have ASCII format. The stem of the output file names is defined by the file_outputname input entry. Assuming file_outputname = filter, the following files are created:
- filter.log:
- contains information on the parameters used in the filter-diagonalization as well as a detailed list of the results
- filter.eig:
- contains exclusively those results of the filter-diagonalization procedure that have not been labeled spurious
- filter.inp:
- contains a compilation of all input or default parameters
If file_fourier = yes, the file
- filter.ft:
- contains the Fourier transform of the autocorrelation function
is also created.
The information given in the log file can be grouped into two parts. In the first part the calculation that has been made is characterized. The amount of memory required is given and all relevant definitions from the input file are recapitulated. In the second part the results are displayed. Some statistical quantities such as the number of accurate states found and the mean and extremal errors are given. The energies, intensities, and errors are then presented in a table. The energies are given in the same unit as the energies in the input file. If specified, the differences between the computed results and the values from The Comparison File are also shown. Here an example log file can be found.
The output of the complex symmetric version, fdc, is mainly the same as of the real symmetric version, filter, except that also the widths are output. We write the eigenvalues of the complex symmetric Hamiltonian in the form E - i W/2. The results displayed are then E and W, in the unit system defined in the input file.
In the eigenenergy file those results of the filter-diagonalization procedure that are considered accurate are compiled. The file has two columns. The first one contains the computed energy (given in the same unit as the energies in the input file) and the second one the corresponding intensity:
-0.6157956320071E+001 0.1490531840265E-001This file can be used for instance by Plot Routines to plot the spectrum. It can also be specified as Comparison File in the input file.
-0.6156696223016E+001 0.6947238058296E-002
-0.6152255257511E+001 0.1894580215085E-002
-0.6150741128375E+001 0.6643072672641E-002
-0.6146348651616E+001 0.2202109986605E-002
-0.6143928233993E+001 0.6573779754760E-003
-0.6139563921858E+001 0.3257708374273E-002
-0.6137741436952E+001 0.9664705874401E-003
-0.6135453802428E+001 0.1772057222989E-003
For analysing the eigenvalues, and in particular for comparing the results of different calculations, a couple of programs exist, which are detailed in Auxiliary Programs.
In the case of a complex symmetric Hamiltonian, the file has an additional column in the middle, containing the widths W (as defined above):
-6.1579563200E+000 8.8434010961E-002 1.4905318402E-002The energies and widths are given in the same unit as the energies in the input file. The eigenenergy file may again be specified as Comparison File in the input file.
-6.1566962230E+000 9.7385842510E-002 6.9472380582E-003
-6.1522552575E+000 3.4969316220E-002 1.8945802150E-003
-6.1507411283E+000 5.0983233635E-002 6.6430726726E-003
-6.1463486516E+000 2.5467053413E-002 2.2021099866E-003
-6.1439282339E+000 6.4363025276E-002 6.5737797547E-004
-6.1395639218E+000 6.6497841620E-002 3.2577083742E-003
-6.1377414369E+000 4.4885779929E-003 9.6647058744E-004
-6.1354538024E+000 9.3516450317E-002 1.7720572229E-004
This file contains a complete list of the input or default parameters. Below an example is given.
# Keyword Parameter Status prop_outputinterval = 0.200000E-02 # computed prop_timeconversion = 1.0d0 # input prop_energyshift = 0.0d0 # input prop_sodintegrator = no # input prop_skipauto = 1 # input prop_wavepackets = 000004 # computed prop_usepackets = 4 # cmdline file_autopath = . # default file_autoformat = mctdh # input file_psilength = 0 # default/ignored file_exactname = # default file_outputname = filter_c0_p4 # cmdline file_fourier = no # input file_fourierresolution = 512 # input/ignored window_minimumenergy = 50.0d0 # compulsory window_energywidth = 40.0d0 # compulsory window_energypoints = 25 # compulsory window_energyconversion = 1.0d0 # input filter_timeinterval = 2.0d0 # cmdline filter_function = cos # input filter_gausswidth = 0.3d0 # input/ignored filter_autoorder = zero # cmdline filter_cutoff = 1.0d-10 # input vp_principle = H # input vp_energyshift = 0.700000E+02 # default/ignored vp_iterative = no # input/ignored vp_accuracy = 1.0d-9 # input/ignored error_estimate = variation # input error_spurious = 0.1d0 # input/ignored error_principle = 1/H # input error_energyshift = 0.700000E+02 # default error_iterative = no # input/ignoredDisplayed are the keywords and the corresponding parameters. The status information in the last column has the following meaning:
- compulsory:
- the corresponding input entry is compulsory
- computed:
- the corresponding parameter was computed, e.g. from the data in the autocorrelation files
- default:
- the default parameter was used
- input:
- the parameter was taken from the input file
- cmdline:
- the parameter was taken from the command line
- default/ignored:
- the given parameter is a default value but was ignored, i.e. never used in the calculation
- input/ignored:
- the entry was present in the input file but ignored
- cmdline/ignored:
- the entry was present in the command line but ignored
An input entry is ignored if it is relevant only in conjunction with another parameter which has not been chosen in the input file. (E.g. filter_gausswidth will be ignored unless filter_function = gauss.) Note that the file strictly obeys the input syntax (see Input Structure) and hence can be used as input file for the filter program.
The fourier file contains the Fourier transform of the (zero-order) autocorrelation function, folded with the selected damping function, in the specified energy range. The two columns represent the energy and intensity:
-6.160000000000000 260.6785668291813The energies are given in the same unit as the energies in the input file. With the aid of Plot Routines the Fourier transform can be plotted.
-6.159863013698630 277.5184621569997
-6.159726027397260 294.4093306381895
-6.159589041095891 311.2240400316336
-6.159452054794521 327.8298286201650
-6.159315068493151 344.0899576666821
-6.159178082191780 359.8655338013862
-6.159041095890410 375.0174696607140
-6.158904109589042 389.4085459223212
-6.158767123287672 402.9055335833113
...
When the program fdc is used, the Fourier transform of the autocorrelation function is a complex function of the complex argument E - i W/2. The fourier file hence contains the coordinates E and W, as well as the real part, imaginary part, and absolute value of the Fourier transform, e.g.
1.29032E-01 -7.74194E-01 2.47395E-03 -1.44438E-03 2.86473E-03The range of both E and W is the energy window defined in the input file.
1.29032E-01 -8.38710E-01 2.39061E-03 -1.37435E-03 2.75750E-03
1.29032E-01 -9.03226E-01 2.31519E-03 -1.30643E-03 2.65836E-03
1.29032E-01 -9.67742E-01 2.24602E-03 -1.24107E-03 2.56610E-03
1.29032E-01 -1.03226E+00 2.18188E-03 -1.17853E-03 2.47983E-03
1.29032E-01 -1.09677E+00 2.12189E-03 -1.11894E-03 2.39884E-03
1.29032E-01 -1.16129E+00 2.06537E-03 -1.06236E-03 2.32258E-03
1.29032E-01 -1.22581E+00 2.01184E-03 -1.00876E-03 2.25058E-03
1.29032E-01 -1.29032E+00 1.96092E-03 -9.58098E-04 2.18247E-03
The computed results (so far of the real symmetric program only) can be visualized by shell-scripts that make use of the Gnuplot program. The existing scripts are tabulated below.
Script | Required Files | Description |
---|---|---|
plfdspec | file_outputname.eig | Plots the line spectrum obtained from the filter-diagonalization procedure. |
plfdfour | file_outputname.ft | Displays the Fourier transform of the autocorrelation function. |
The shell-scripts recognize a number of arguments. Type
plfdspec -h
or
plfdfour -h
to get an overview.
An auxiliary program exists which is designed for comparing the results of different calculations obtained with the real filter program. (It does not work for the complex one since it assumes that the results are lists of energies and intensities, and not energies, widths, and intensities.) It is called as
fdmatchver k file_1 ... file_k | sort -n | less
where ver is the version number, k specifies the number of files to be compared, and file_1 ... file_k are the names of the files. The files are typically the *.eig files generated by the filter program.
For instance, if file_1 contains the five energies and intensities
1285.3673185311E+000 1.0740277960000000E-001and file_2 contains the six pairs of values
1371.0602365654E+000 4.3956966964161830E-004
1388.2371505938E+000 3.4935267560000000E-002
2349.1285061750E+000 3.1545981720000000E-002
2548.3232400815E+000 1.0620001100000000E-001
1285.3668549833E+000 1.074020343487891E-001then type
1388.2368347075E+000 3.491722659565360E-002
2349.1269376485E+000 3.153875507967670E-002
2383.3355167369E+000 1.263746447807894E-005
2498.4605910470E+000 5.299441725000872E-005
2548.3225966500E+000 1.061952361250140E-001
to get the combined list
fdmatch 2 file_1 file_2 | sort -n | less
1285.367318531 1.074027796E-01 1285.366854983 1.074020343E-01Where no matches between energies have been found, the corresponding entry is left blank.
1371.060236565 4.395696696E-04
1388.237150594 3.493526756E-02 1388.236834708 3.491722660E-02
2349.128506175 3.154598172E-02 2349.126937649 3.153875508E-02
2383.335516737 1.263746448E-05
2498.460591047 5.299441725E-05
2548.323240081 1.062000110E-01 2548.322596650 1.061952361E-01
A second auxiliary program named fdcheck analysis the output of fdmatch, i.e. it checks whether a state has been found in all comparison calculations and computes the internal errors. It is called as
fdcheckver k j
where ver is the version number, k denotes the number of data sets (a set consists of two numbers, energy and intensity), and j is a flag which specifies whether (j = 1) or not (j = 0) the first data set is an "exact" one. Such an "exact" data set is ignored in the decision whether a state has been found and in the computation of the internal error. The "exact" data set must be the first one.
For example, with file_1 and file_2 as above, the command line
yields the output
fdmatch 2 file_1 file_2 | sort -n | fdcheck 2 0
1285.367087 0.000232 1.07402E-01 3.72650E-07Only those energies being present in all data sets are put out. The first column gives the (mean) energy, the second one the internal error (i.e. the standard deviation), the third one the (mean) intensity, and the last one again the corresponding internal error.
1388.236993 0.000158 3.49262E-02 9.02048E-06
2349.127722 0.000784 3.15424E-02 3.61332E-06
2548.322918 0.000322 1.06198E-01 2.38745E-06
If there was also an exact data set
1285.0000000000E+000 1.1000000000000000E-001stored in file_0, then type
1388.0000000000E+000 3.5000000000000000E-002
2548.0000000000E+000 1.1000000000000000E-001
3000.0000000000E+000 1.0000000000000000E-003
to obtain the result
fdmatch 3 file_0 file_1 file_2 | sort -n | fdcheck 3 1 | less
1285.000 1285.367 0.000 0.367 1.10E-01 1.07E-01 3.73E-07 2.60E-03The first column contains the "exact" energy (of file_0), the second the mean energy (of file_1 and file_2 only), the third the standard deviation or internal error (of file_1 and file_2 only), and the fourth the absolute difference between the first two columns, i.e. the "exact" error. The next four columns contain the corresponding values for the intensities. In the example, the value in line three is spurious and that in the last line is missing.
1388.000 1388.237 0.000 0.237 3.50E-02 3.49E-02 9.02E-06 7.38E-05
2349.128 0.001 3.15E-02 3.61E-06
2548.000 2548.323 0.000 0.323 1.10E-01 1.06E-01 2.39E-06 3.80E-03
3000.000 1.00E-03