| |
|
|
|
The output from a Jaguar run always includes a Jaguar output file, which contains the primary output, and a log file, which is mainly useful as a job summary as the job is being run. If you request other output options from the Files window, whose button is found under the Output heading, various other files can also be generated as output.
This chapter begins with a description of the Jaguar output file for a standard Hartree-Fock calculation, continuing with a discussion of the changes in the output for various other calculation options and the output options which can be set from the Standard, Files, Per Iteration, and Orbitals windows found under the Output heading. The final section explains the log file, which is the file displayed when the Check Job button in the interface's file viewer window is clicked as a job runs.
Throughout this chapter, footnotes indicate the Jaguar input file keywords and sections that correspond to particular interface settings. If you are working from the interface, you can ignore these footnotes, but you may find them helpful if you decide to use input files to submit jobs without using the interface.
If you can run perl scripts on your system, you can obtain summaries of Jaguar results in simple table form by using the command "jaguar results". Jaguar will then search output files for the information you request through "jaguar results" options.
The tables produced by "jaguar results" can describe results from one job or several jobs. The results can be restricted to final results from each job listed (the default), or can include intermediate results (SCF energies for each geometry in an optimization, for instance). By default, each line lists information that pertains to the entire input structure, but you can also request some kinds of information for each individual atom in the structure. Each of these types of "jaguar results" tables are described below.
You can use "jaguar results" tables to summarize key output in an easy-to-read format or to produce files in a particular format for use in other scripts or programs you write. For either of these purposes, we recommend creating aliases for the "jaguar results" commands you find most useful, and, if you want, writing the "jaguar results" output to a file for later viewing or use. If you need help accomplishing either of these tasks, please contact Schrödinger or your system administrator.
By default, each row of the "jaguar results" table (except the title row) corresponds to the final results from a Jaguar output file that was listed in the "jaguar results" command. For instance, if you entered the command
from a directory containing the output files RuCp2.out and piperidine.out, you would get a very simple table like this:
where the first line lists the final energy from the job RuCp2 and the second lists the final energy from the job piperidine.
If your "jaguar results" command includes the flag "title", the table will have column headings indicating the type of information listed. For instance, if you modified the command above to include the "title" flag:
your "jaguar results" table would look like this:
(If you want to see ahead of time what the column headings of your table would look like without any results listed, use the "titleonly" flag.)
The "jaguar results" tables can list both information describing the job run (for instance, its name, the basis set and SCF method used, or the stoichiometry of the molecule) and information about the results of the job (for example, the final energy or dipole moment). Each of these types of information appears in a column in the table. If you use the "-title" flag, column labels will indicate the type of information in each column. The columns appear in the table in the same order they are listed in the "jaguar results" command.
Table 5.1.1 and Table 5.1.2 show some options you can use to get "jaguar results" tables summarizing final results of a job. Table 5.1.1 shows options you can use to get a description of the job run (as determined by your input file); Table 5.1.2 lists options that help you obtain actual final calculated properties and results.
The order of the list of "jaguar results" options determines the order of the columns of information in the table (from left to right). For instance, the command
(where h2o.out and h2o_b3lyp.out are output files from jobs at the Hartree-Fock and density functional theory (with the B3LYP functional) levels, respectively) gives the "jaguar results" table
with the job name, method, and energy listed from left to right in the same order they were in the "jaguar results" command.
By default, only the final results are reported for each job; therefore, for instance, a table of results from three jobs would have three rows of information. However, you can also request that information from each geometry, SCF, or gradient calculation be reported in a different row of the "jaguar results" table. For instance, the command
here produces a table showing the convergence of a BLYP geometry optimization of water:
lists the "jaguar results" options that let you specify when to report intermediate (and final) results from jobs. The "all" option, which lets you track the progress of a geometry or transition state optimization, is likely to be the most useful of the options shown in Table 5.1.3. The "allscf" option can be used for intermediate results in complex non-optimizations, such as solvation jobs.
You can use the options from Table 5.1.3 in combination with any of the "jaguar results" options described earlier, such as job name or energy results (see Table 5.1.1 and Table 5.1.2). You can also report various types of output that are primarily useful for reporting for the intermediate steps of the job, as shown in Table 5.1.4
. (You can use the options in Table 5.1.4 without any of the flags shown in Table 5.1.3-that is, to get final results-but they are generally most useful for judging convergence, particularly of geometry optimizations.)
As described above, by default, each line of output from a "jaguar results" command lists information that pertains to the entire input structure, but you can also request some kinds of information for each individual atom in the structure.
lists the "jaguar results" options that let you print tables of coordinates, forces, or charges for individual atoms. You can use these options in combination with the options from Table 5.1.1, which describe the job run, or with the options from Table 5.1.3, which request output for different stages of a job. However, you should not use the atom-related "jaguar results" with any of the options that request information pertaining to the entire molecule (the "energy" option, for instance).
The contents of a Jaguar output file vary according to the calculation and output selections made. This section describes the output file for a standard, default, single point, closed shell Hartree-Fock calculation. Section 5.3 describes the variations in the output file for the calculation options described in Chapter 3.
All output files begin with a line listing the job name, the machine upon which the job ran, and the time the job was started, followed by the general copyright information for the version of Jaguar which was used for the run. The rest of this section describes output from each individual Jaguar program run for a default calculation.
The output from the program pre begins with a description of the calculation to be performed: its job name, the directory containing the executables used to run the job, the directory containing the temporary files, comments from the input file (if any), and the names and paths of any non-default data files used for the calculation (as explained in section 8.2 and Chapter 9). Comments from the input file include any text entered in the Comment box in the Run or Save window, as well as a comment about the point group if the geometry was symmetrized as described in section .
Next, the basis set used for the calculation, the molecule's net charge and multiplicity, and the number of basis functions used for the calculation are specified. This information is followed by the molecular geometry input, which gives the atom label and coordinates for each atom. (If the atom labels provided in the geometry are not unique-for instance, if two hydrogens are each called `h'-this information is preceded by a list of original atom labels and new atom labels assigned by the program.)
The molecule's symmetry is analyzed, a process which may involve translating and rotating the molecule. These procedures are noted in the output file, along with the point group used for the calculation, the nuclear repulsion energy, and the symmetrized geometry, which is used for the rest of the calculation.
One-electron integrals are calculated by the onee program, which outputs the smallest eigenvalue of the overlap matrix S and the number of canonical orbitals used for the calculation. Canonical orbital eigenvectors with very small eigenvalues (less than 5.0 x 10-4) are removed and thus are not counted. The eigenvalue cutoff can be controlled by setting the keywords cut10 and cut20 to the desired value in the gen section of the input file.
The program hfig constructs a starting wavefunction (initial guess) for a Hartree-Fock calculation. The output from the program hfig for a default calculation begins with the line, "initial wavefunction generated automatically from atomic wavefunctions." Next, a table lists the number of orbitals, and of occupied orbitals in each shell, having each irreducible representation for the appropriate point group. Finally, the orbital occupation for each shell is listed; an occupation of `1.000' indicates a closed shell. An example, for a calculation of water using a 631G** basis set, follows:
The probe program, which follows hfig and insures orthogonalization, has no significant output.
The output for the grid generation done by the program grid lists the number of grid points for each atom, as well as the total number of grid points, for each grid used in the application of the pseudospectral method. If you would like more information about these grids, see section 9.4. The rwr program, which generates the Q operators needed for the pseudospectral method, runs next, but has no significant output.
An example of the next output, from the program scf, is given here, again for a water molecule, and will be explained below.
The output from the program scf begins with a list of information detailing the number of electrons in the molecule, the number of alpha and beta electrons, the total number of orbitals for the calculation, the numbers of core, open shell, occupied, and virtual orbitals, the number of Hamiltonians used for the calculation, the number of shells, and the calculation type.
Next, the energy output from the SCF iterations is shown in table form. Some of the text for the column headings should be read down rather than across. The number of the iteration is provided first in each row, followed by a "Y" or "N" indicating whether the Fock matrix was updated or not. When the Fock matrix is updated, the changes are made using a differential density matrix whose elements simply reflect the changes in the density matrix elements from the previous iteration to the current one.
The next entry indicates whether the DIIS convergence scheme was used for that iteration. As above, "Y" or "N" indicate yes or no. The DIIS method produces a new estimate of the Fock matrix as a linear combination of previous Fock matrices, including the one calculated during that iteration. DIIS, which is enabled by default, usually starts on the second iteration, and is not used on the final iteration. If the entry in this column reads "A," it indicates that DIIS was not used for that iteration, but the density matrix was averaged.
The cutoff set for each iteration is indicated under the "icut" heading. Cutoff sets are explained in the cutoff file description in section 9.5.
The grid column lists the grid used for that iteration, which must be one of the grid types coarse (signified by a C), medium (M), fine (F), or ultrafine (U). See the subsection Grid and Dealiasing Function Keywords in section 8.6, and the description of the grid file in section 9.4, if you want more information on grids and grid types.
The total energy for the molecule in Hartrees appears in the next column, followed by the energy change, which is the difference in energy from the previous iteration to the current one.
The RMS density change column provides the root mean square of the change in density matrix elements from the previous iteration to the current one.
In the last column, the maximum DIIS errors listed provide a measure of convergence by listing the maximum element of the DIIS error vector. For HF and DFT closed shell calculations, the DIIS error vector is given by FDS - SDF in atomic orbital space, where F, D, and S are the Fock, density, and overlap matrices, respectively. For open shell and GVB cases, the definition of the error vector is given in reference [11].
After the energy information for each SCF iteration is provided, a summary of the components of the final, converged energy is listed. The nuclear repulsion, one-electron, two-electron, and electronic contributions are all listed, followed by the total. Each of these energies is labeled with a letter (for example, `A' for the nuclear repulsion), and information to the right of some of the energies describes the relations between the components in terms of these letters. A line beneath the table summarizes the calculation type and energy, as well as the number of SCF iterations.
(If the input system's spin multiplicity is not singlet, the default, the summary of the SCF output also includes a breakdown of the two-electron contribution to the energy into Coulomb and exchange parts. For each of these parts, the contribution from each Hamiltonian is listed.)
The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energies are listed next. Finally, the energies for each occupied orbital and for the ten lowest-energy virtual orbitals (if that many exist) are provided, with each orbital identified by a symmetry label. Virtual orbitals and eigenvalues are determined in the same manner as in ref. [89]. The virtual orbitals are obtained by diagonalizing
, where f is the occupation of each orbital (1 for a closed shell). (For closed shell Hartree-Fock calculations, this definition yields the standard orbitals and eigenvalues.)
Finally, the cpu time for the job, the machine upon which the job ran, and its time of completion are noted at the end of the output file.
Any time you make a non-default setting for a calculation, the output from the program pre will note the non-default options chosen. This output will appear above the molecular geometry output from the pre program. This section describes the changes in output for various calculation settings described in Chapter 3.
Generally, only the format changes that result from these settings are discussed below! Naturally, these settings will often change the data listed. For information on the settings themselves, see Chapter 3. Options which have no significant impact on the output format are not discussed in this section.
If you use density functional theory for the SCF calculation, the output above the SCF table lists the functional or combination of functionals used. The energy information for DFT calculations includes the breakdown of the two-electron energy into Coulomb and exchange-correlation terms. For DFT calculations, virtual orbitals are obtained by diagonalizing
, where f is the occupation of each orbital (1 for a closed shell). (For closed shell calculations, this definition yields the standard orbitals and eigenvalues.)
The scf output from post-SCF DFT energy evaluations (GVB-DFT calculations, for instance) first lists the standard output for the HF, GVB, or DFT SCF calculation, then lists the energy breakdown and total energy from the post-SCF DFT analysis. Since the post-SCF DFT treatment does not change the wavefunction, no orbital output is reported from this step.
The output from the program pre for non-default options contains the detailed description of customized functional combinations for SCF or post-SCF DFT calculations.
If you perform a local MP2 calculation, the output from the programs pre and hfig is somewhat different than for a Hartree-Fock calculation, since the use of symmetry is turned off automatically for LMP2 calculations. The output from the program scf includes the Coulomb and exchange contributions to the two-electron terms for these calculations, and the symmetry labels are not included in the output of orbital energies.
The program loclmp2, which computes localized orbitals, runs after scf in an LMP2 calculation, and its output notes the number of orbitals that are localized. Below that output, the output from the program lmp2 appears.
For local local MP2 calculations, the output begins by listing the localized orbitals involved in the local local MP2 treatment-namely, the localized orbitals centered on one or both atoms in the pairs of atoms for which an LMP2-level treatment was requested.
All LMP2 output includes a description of the type of orbitals used in the MP2 calculation. First, it lists the total number of orbitals. Next, it lists the number of frozen core and valence MP2 orbitals. The numbers of core and valence orbitals will be affected by your choice from the LMP2 window of whether to use valence electrons only or all electrons for the atoms in the calculation. Next, the numbers of occupied and virtual orbitals for the molecule are listed. The list ends with the number of exchange Hamiltonians.
Some information on the convergence of the LMP2 energy correction appears below the list of orbital information, followed by the Hartree-Fock energy and the LMP2 energy correction, which gives the improvement to the energy over the HF value. The total LMP2 energy (the HF energy plus the correction) is given immediately afterwards. (If your job is a local local MP2 calculation and you want to see the energy from each LMP2 pair, use the gen section keyword setting ip170 = 2, as described in section 8.6 under Standard Output Keywords.)
If a GVB calculation is performed from a Hartree-Fock initial guess, the pre program output lists a table of GVB pair information below the list of non-default options. The information in the table includes whether a restricted configuration interaction (RCI) calculation including that pair will be performed (Y or N for yes or no), and what the configuration interaction (CI) coefficients are for the pair. Since the use of symmetry is turned off automatically for GVB calculations, the output from the programs pre and hfig is somewhat different than for a Hartree-Fock calculation. Also, the program gvbig runs after hfig, if the GVB initial guess is being generated from the HF initial guess.
The output from the scf program is more extensive than for a default HF calculation. First, the number of GVB pairs and the number of GVB orbitals are added to the list of electron and orbital information preceding the table of SCF iteration information. Secondly, the summary of the SCF output is followed by a breakdown of the two-electron contribution to the energy into Coulomb and exchange parts. For each of these parts, the contribution from each GVB Hamiltonian is listed. After this information, the intra-pair exchange energies and their sum are listed. Finally, a table of GVB pair information is given. Here is an example of this GVB information in the SCF output, for a water molecule with two GVB sigma pairs:
Each row in the GVB pair information table lists the pair number, the orbital number (after all core and open orbitals have been assigned numbers), Hamiltonian number (after the core Hamiltonian and any open Hamiltonians have been assigned numbers), and shell number (after the core shell and open shell, if they exist, have each been assigned a number) corresponding to each natural orbital, and CI coefficient corresponding to each GVB natural orbital in the pair. Next, the overlap between the two corresponding non-orthogonal orbitals for that pair is listed, followed by the CI energy lowering, which is a guide to the energy change resulting from the inclusion of the second natural orbital in the calculation.
If a GVB calculation is performed from a Hartree-Fock converged wavefunction, the program scf runs twice, once to obtain the HF converged wavefunction, and once to perform the final GVB calculation. The SCF output from the first scf run will look like the scf output from a standard HF calculation; the output from the second run will have the format described above for a GVB calculation from an HF initial guess.
For restricted configuration interaction calculations, the SCF output is the same as for non-RCI GVB calculations, but the output from the program rci appears after the SCF output. The RCI output first lists information on the total number of orbitals, the number of core orbitals for the RCI calculation, the numbers of open shell and GVB orbitals, the number of GVB and RCI pairs, the numbers of occupied and virtual orbitals, the numbers of Coulomb and exchange Hamiltonians, and the multiplicity. Next, the total energy of the GVB wavefunction, which was obtained from the SCF procedure earlier, is listed and broken down into a nuclear repulsion term and terms from the core, which is treated with Hartree-Fock methods, the GVB pairs, and the open shell contribution. If you specified GVB pairs that were not also RCI pairs, a non-zero value is listed for the non-RCI GVB pair energy. The number of RCI spatial configurations and the number of RCI configuration state functions follow. Each RCI configuration state function is the product of a contracted spatial function and a spin function; the number of these functions indicates the size of the RCI expansion.
The rci output lists the RCI initial guess at the energy next, followed by the output from the calculation of the converged total RCI energy. For small cases, the initial guess and the converged energy may agree exactly, since the RCI coefficients are obtained by one diagonalization of a small matrix. For larger cases, the output includes the results of the iterative diagonalization.
The output format description for optimizations in this subsection applies to calculations of either minimum-energy structures or transition states. Although the Hessians used during these calculations are different, the Jaguar programs run are the same, and the output format is very similar. (Exceptions are described below.)
If you calculate an optimized molecular structure, a transition state, or forces, any SCF calculations during the run use the RMS density change convergence criterion described in section 3.9 instead of the usual energy convergence criterion. Therefore, these SCF calculations often proceed for several more iterations than single point energy calculations yield.
If you select forces only for the Optimize geometry setting, the programs der1a, rwr, and der1b will run after scf does. The forces felt by each atom in the unoptimized geometry will be output from der1b, in a table listing each atom and the components of the force upon it in the x, y, and z directions. The x, y, and z components of the total force on the molecule are listed in the last line, and provide a judge of how accurate the force calculations are in most cases, since they should generally be zero. An example of this force table for a water molecule optimization follows:
When force calculations or optimizations of a system's minimum energy structure or transition state are performed at the LMP2 level, the program der1b never runs. Instead, forces are calculated by the programs lmp2der, lmp2gda, and lmp2gdb. The last of these programs provides a table of output listing the forces on each atom in the same format as the sample table above.
If Optimize geometry is set to minimum energy or transition state, Jaguar prints bond length and angle information in the output from the program pre. This information appears in the output file because the bond lengths and angles option from the Standard window under the Output heading is automatically enabled for geometry optimizations. If you have constrained certain bond lengths or angles of the geometry so that they are frozen during the optimization, as described in section 4.2, the constraints are also listed in the pre output.
At the end of the first SCF calculation, the programs der1a, rwr, and der1b run, calculating the forces felt by each atom in the unoptimized geometry and writing them to the output file, as described above.
These force results are followed by the output from the program geopt, which includes a number indicating how many times it has been called (for example, 1 for the first iteration) in the "start of program geopt" line. Every time geopt is called, this number is updated. However, since geopt can be called for Hessian refinement steps as well as for generating new geometries during an optimization, and since geometry optimizations occasionally revert back to a previous geometry and "restart" the calculation from there, the next line of the geopt output reports what sort of step is being performed and numbers that step accordingly.
If the program detects that the input lists separate fragments, each of which contain only atoms unbonded to the atoms in any other fragment, as for a van der Waals complex, then the number of fragments is listed near the start of the geopt output.
For transition state optimizations, the eigenvalues of the nuclear Hessian are reported the first time geopt runs. If the initial Hessian is being refined, the coordinates for the refinement and their eigenvalues are listed. (If a coordinate you have specified is inappropriate because of symmetry restrictions or other constraints, the output will indicate the problem.) The geopt output then lists information on the current (original) geometry's gradient elements, describes the small step it will use to alter the first coordinate used in the Hessian refinement, describes the internal coordinates and optimization variables as stretches, bends, or torsions, and indicates how it generates a new geometry by altering the relevant coordinate by the amount described by the step size.
The new geometry generated for Hessian refinement is then used to obtain energy and gradient information, a process that requires the programs onee, grid, and rwr to run and generate output in the usual formats, which is followed by output from the program scf, which now starts with the calculation type and the table showing the energy output from each SCF iteration (skipping the listed information about electrons, orbitals, and so on). The output further continues with output in the usual formats from der1a, rwr, and der1b. The information obtained on that geometry is then used in geopt, which runs a second time, reporting similar information about the planned changes to the molecular structure for the next Hessian refinement step (if there is one) and reporting the change in total energy from the original geometry to the geometry for the first Hessian refinement step as well. This process of altering single coordinates from the original geometry and calculating energies and gradients for the changed geometry continues until all requested Hessian refinement steps have been performed, which the output indicates with a line beginning "Hessian optimization completed." After that point, geopt performs a geometry optimization step from the original geometry, and the actual optimization continues until convergence.
For transition state optimizations, the output for iterations that follow any Hessian refinement includes information identifying the transition vector used for that iteration. This output includes the transition vector's eigenvalue and the stretches, bends, or torsions that are its most important components.
For any optimization iteration using level shifting, after any relevant lines of geopt output described above, some information on the computed level shift (which may then be adjusted to satisfy step-size constraints) is included in the output. For optimization steps past the first geometry change, the change in total energy from the previous geometry to the newly calculated geometry (in Hartrees) is listed next.
The geopt output then notes the maximum element of the analytic gradient calculated by the earlier programs; the root mean square of the gradient elements; the step size predicted for the geometry change, the trust radius for that iteration and, if it is smaller than the step size, the factor used to scale the step size so it is no larger than the trust radius; the maximum element of nuclear displacement; and the root mean square element of the nuclear displacement. The predicted energy change for the new structure generated by geopt is also listed.
The values for the energy change, gradient, and nuclear displacement described in the previous paragraph are important because they are each tested against the convergence criteria determined by the Convergence criteria setting from the interface's Optimization window, as described in section 4.1, or, alternatively, the criteria set by the gconv keywords in the input file. The criteria are described in detail in the Geometry Optimization and Transition State Keywords subsection of section 8.6. If the gradients are converged and the energy change is below 2.5 x 10-7, the optimization will stop (unless it is on the first geometry optimization iteration). Similarly, if the gradients are converged and one of the gradient criteria is 5 times lower than the convergence level, then the optimization will stop if the energy change is less than 2.5 x 10-6.
The symbol following each quantity used to judge convergence indicates how well converged it is. The symbol `.' indicates convergence criteria that are not satisfied, `*' indicates criteria that are satisfied, `#' indicates criteria that are quite well satisfied, `!' indicates values that are essentially zero. If the convergence criteria mentioned are not sufficiently well met, and if the geometry optimization has not already gone through as many iterations as the Maximum iterations setting from the Optimization window, the output will note "molecular structure not yet converged..." and the optimization will continue.
The output next lists the movement of the center of mass. If the output option for the bond length and angles is enabled, the output then lists this information for the new structure. Finally, the nuclear repulsion energy for the new geometry is listed.
If the molecular structure was not yet converged and the maximum number of geometry optimization iterations allowed was not reached in the previous iteration, the output from more geometry optimization iterations will follow. The output from each iteration begins with onee, grid, and rwr output in the usual formats, and continues with output from scf, which now starts with the calculation type and the table showing the energy output from each SCF iteration (skipping the listed information about electrons, orbitals, and so on). The output further continues with output in the usual formats from der1a, rwr, and der1b, winding up with the output from geopt. The last such geometry optimization iteration contains, in the geopt output, either the line, "Geometry optimization complete," or the line, "stopping optimization: maximum number of iterations reached," depending on whether the convergence criteria were met before the maximum number of iterations was reached.
Geometry or transition state optimizations using GVB-RCI run in much the same way as described above for HF, GVB, or DFT optimizations, except that the forces for the optimization are computed numerically rather than analytically. Consequently, the der1a and der1b programs never run; instead, when forces are needed, the structure's energy is evaluated at 6Natom perturbed geometries, where Natom is the number of atoms, and the forces are computed numerically. The program nude generates each perturbed geometry by moving an atom a small amount in the positive or negative x, y, or z direction, and also evaluates the numerical derivatives when calculations on all perturbed geometries are complete, listing them in the output in a force table similar to the usual geometry optimization force table described for HF, GVB, or DFT systems. The program geopt still runs in the usual way as well, computing each iteration's new geometry using the available forces.
Performing a solvation calculation involves several iterations in which the wavefunctions for the molecule in the gas phase are calculated; the program ch performs electrostatic potential fitting which represents the wavefunction as a set of point charges on the atomic centers; the interactions between the molecule and the solvent are evaluated by Jaguar's Poisson-Boltzmann solver, which fits the field produced by the solvent dielectric continuum to another set of point charges; these solvent point charges are passed back to scf, which performs a new calculation of the wavefunction for the molecule in the field produced by the solvent point charges; ESP fitting is performed on the new wavefunction; the solvent-molecule interactions are reevaluated by the Poisson-Boltzmann solver; and so on, until the solvation energy for the molecule in the solvent converges.
For solvation calculations on neutrally charged systems in water whose atoms all have atomic numbers under 19 (HAr), by default, the program pre evaluates the Lewis dot structure for the molecule or system and assigns atomic van der Waals radii accordingly. (For more information on this process, see section 9.6.) These van der Waals radii are used to form the boundary between the solvent dielectric continuum and the solute molecule. The Lewis dot structure and van der Waals radii information both appear in the output from the program pre. The radii are listed under the heading "vdw2" in the table of atomic information below the listing of non-default options. See section 8.9, which describes the atomic section of the input file, if you want information on the other information in this table.
After the pre output, the usual output appears for the first, gas-phase calculation, except that the energy breakdown for the scf output also describes the electron-nuclear and kinetic contributions to the total one-electron terms in the energy, as well as the term V/T (where V is the potential energy and T is the kinetic energy), which indicates how well the calculation is agreeing with the virial theorem (it should be 2.00).
After the first scf output, the output from the first run of the program ch appears. Since performing a solvation calculation enables ESP fitting to atomic centers, the usual output for that option, which is described in the subsection Properties later in this section, is included every time output from the program ch appears in the output file. The post program writes out the necessary input files for the Poisson-Boltzmann solver; this step is noted in the output file.
The next output section comes from the Poisson-Boltzmann solver. The output includes information on the area (in Å2) of the molecular surface formed from the intersection of spheres with the van der Waals radii centered on the various atoms; the reaction field energy in kT (where T = 298 K), which is the energy of the interaction of the atom-centered charges with the solvent; the solvent-accessible surface area (in Å2), which reflects the surface formed from the points whose closest distance from the molecular surface is equal to the probe radius of the solvent; and the cavity energy in kT, which is computed to be the solvation energy of a nonpolar solute whose size and shape are the same as those of the actual solute molecule, as described in reference [15].
The output from the Jaguar program solv follows the Poisson-Boltzmann solver results, giving the number of point charges provided by the solver to model the solvent, the sum of the surface charges, the nuclear repulsion energy already calculated by Jaguar, the nuclear-point charge energy representing the energy of interaction between the molecule's nuclei and the solvent point charges, and the point-charge repulsion energy, which is calculated but not used by the rest of Jaguar because it is irrelevant to the desired solvation results.
After this output, the output for the second solvation iteration begins. The output from scf comes first, giving the results for the molecule-and-solvent-point-charges system. An example, from the first solute-with-solvent-point-charges scf run in a calculation of 631G** water in cyclohexane, using the Jaguar solver, is given here:
As for any later solvation iterations, the scf output begins with the calculation type and the table of energy results for each iteration, skipping the list of information about the molecule's electrons and orbitals, and the energy information below the table includes several additional terms, whose relations to each other are described with the usual alphabetic labels. First, the total of the terms with no electron contribution is listed (term (A)), followed by terms (B) and (C), the nuclear-nuclear and nuclear-solvent energies.
Next, the total one-electron energy is listed, along with its three components, the electron-nuclear, electron-solvent, and kinetic energies. The total two-electron energy, and the total of the one- and two-electron energy terms, the electronic energy, follow. Term (N), the total of the zero-, one-, and two-electron terms, is then listed, with the label "Total quantum mech. energy." This term corresponds to the final energy from the scf energy table for that iteration, and includes the entire energies for the molecule-solvent interactions.
The output next includes the gas phase and the solution phase energies for the molecule, since these terms are, of course, necessary for solvation energy calculations. The first solution phase energy component is the total solute energy, which includes the nuclear-nuclear, electron-nuclear, kinetic, and two-electron terms, but no terms involving the solvent directly. The second component of the solution phase energy is the total solvent energy, which is computed as half of the total of the nuclear-solvent and electron-solvent terms, since some of its effect has already changed the solute energy. Third, a solute cavity term, which computes the solvation energy of a nonpolar solute of identical size and shape to the actual solute molecule, as described in reference [15], is included. The last solution phase energy component (shown only if it is nonzero) is term (T), the first shell correction factor, which depends on the functional groups in the molecule, with atoms near the surface contributing most heavily.
Finally, the list ends with the reorganization energy and the solvation energy. The reorganization energy is the difference between the total solute energy and the gas phase energy, and does not explicitly contain solvent terms. The final solvation energy is calculated as the solution phase energy described above minus the gas phase energy.
The results of the self-consistent reaction field iterations so far performed are summarized after the scf output in the output from the program sole. An example from the final SCRF iteration of water in cyclohexane follows:
The solvation energy is listed in Hartrees and in kcal/mole, and the note beneath it reads either "solvation energy not yet converged..." or "stopping: solvation energy converged," depending on whether the solvation energy has changed by less than the Solvation convergence criterion, which is described in section 3.5. If the solvation energy has converged, the output from the sole program includes a line summarizing the solvation energy iterations and result.
The output from ch and post appears below the sole output. If the solvation energy has converged, the ch output reflects the system's final atomic charges. If the solvation energy has not converged, these charges and the Poisson-Boltzmann solver's files generated by the post program are passed to the solver again, and the solvation iterations continue as previously described, until solvation energy convergence is reached.
Geometry optimizations in solution contain output in the formats described in the previous two subsections, but the optimization output and the solvation calculation output alternates as the calculation proceeds. First, by default, Jaguar computes a gas phase optimized geometry, for which the output is the same as that described above for a standard optimization. Next, the SCRF procedure is used to compute a wavefunction for the solvated system, as for a single point solvation energy calculation. When the solvation energy has converged, Jaguar runs the program pbf once more to get the solvation-related gradient. This pbf output does not contain the usual solvent accessible surface area and cavity energy terms. The programs der1a, dsolv, rwr, and der1b then compute the forces, with the force table in the der1b output in the usual manner, and the program geopt computes the new molecular structure, as usual. For each new structure generated during the optimization, Jaguar first performs the SCRF calculation, then obtains the forces (in solution), and finally generates a new structure. The calculation proceeds until the geometry optimization convergence criteria are reached.
For solvated geometry optimizations, the solvation energy is computed as the difference between the energy of the optimized gas phase structure and the energy of the solvated structure that was optimized in solution.
If you make any non-default selections from the Properties window, the program ch runs and outputs the results to the output file after the SCF iterations, if any.
When multipole moments are calculated, the x-, y-, and zdirection components of the dipole moment and the total magnitude of the dipole moment are reported in Debye, followed by information on any requested higher-order moments and the corresponding traceless higher-order moment tensors. For example, here is the output for a calculation of water's dipole and quadrupole moments:
If ESP charge fitting to atomic centers is performed, the output lists the number of grid points from the charge grid, which is used for the charge fit. It then describes the constraint or constraints for the fit, followed by the calculated atomic charges and their sum. The root mean square error of the charge fitting is also reported; this error is calculated from examining the Coulomb field at each grid point that would result from the fitted charges, and comparing it to the actual field.
If ESP fitting to atomic centers and bond midpoints is performed, the bond midpoints are treated as "dummy atoms" and their descriptions and coordinates are provided before the grid points information. The bond charges from the fit are provided, with the label "bond", along with those on the atomic centers. An example of the output from such a calculation follows, for water:
If the fit is constrained to reproduce the dipole moment (or dipole and higher moments), or any other time both ESP fitting and multipole moment calculations are performed, a new moment (or moments) can be calculated from the fitted charges, as described in section 3.6 in the subsection Electrostatic Potential Fitting. The output from ch begins with the moment or moments calculated for the quantum mechanical wavefunction, in the format for multipole moment calculations. Next, the electrostatic potential fitting information is provided, as described above. Finally, the components and totals of the moment or moments recalculated using the electrostatic potential charges are reported.
If you calculate polarizabilities and first hyperpolarizabilities with the coupled perturbed HF method, the tensor elements in A.U. appear in the output from the program cpolar, which runs after the SCF calculation. Alternatively, if you use the finite field method to calculate the polarizability and/or first hyperpolarizability of the molecule, the output includes data from all the SCF calculations involved. (See section 3.6 for details on the methods used to calculate polarizability and hyperpolarizability.) The data from the program scf includes the term V/T (where V is the potential energy and T is the kinetic energy), which indicates how well the calculation is agreeing with the virial theorem (it should be 2.00). Before each SCF calculation used for the polarizability evaluation, the program polar runs and outputs the electric field (in A.U.) used for the SCF calculation whose output appears immediately afterwards. When all calculations needed for the finite difference method have been performed, the program polar outputs the polarizability tensor in A.U., the first hyperpolarizability tensor in A.U., if it has been calculated, and the dipoles from each SCF calculation, along with information about the electric fields used for the dipole calculations.
If you choose to calculate the electron density, the output from the program elden appears beneath the SCF output. The output lists the number of grid points used for the electron density calculation and the total number of electrons found over the grid. The main output file does not include the charges and grid points for the calculation; that information can be found in the output file jobname.chdens, where jobname.in is the input file for the Jaguar job. The file jobname.chdens lists the Cartesian coordinates and the electron density in A.U., respectively, for each grid point.
If you choose to calculate Mulliken populations by atom, the charge for each atom and the sum of the atomic charges will be noted under the heading "Atomic charges from Mulliken population analysis." If you choose to calculate them by basis function, the atomic charge output will be preceded by a section labeled "Mulliken population for basis functions," listing the atom label, function (labeled with consecutive numbers), type of basis function (S for s, X for px, XX for dxx, etc.), and calculated population. Calculating Mulliken populations by bond yields the populations by atom and basis function as well. An example of this output for a calculation of water using the 631G** basis set is provided below.
You may find it helpful to select the Gaussian function list (basis set) setting from the Standard window, whose button appears under the Output heading, if you wish to have more information about the basis functions. More information on this output option is given in Section 5.4.
If both Mulliken populations and multipole moments are calculated, the multipole moments are calculated from the atomic Mulliken populations as well as directly from the wavefunction, as noted in section 3.6 in the Mulliken Population Analysis subsection. The output lists the multipole moments from the wavefunction, as described earlier; the Mulliken populations, as described just above; and finally the recalculated moments resulting from the Mulliken charges, in the same format used for the earlier moment output.
If you perform an NBO calculation, its output will appear under the heading "Jaguar NBO 4.0."
If you calculate vibrational frequencies by making the appropriate setting in the Frequencies window, any SCF calculations during the run use the RMS density change convergence criterion described in section 3.9 instead of the usual energy convergence criterion. Therefore, these SCF calculations often proceed for several more iterations than single point energy calculations yield.
To compute vibrational frequencies from a Hessian computed during the same job, Jaguar sometimes calculates the second derivatives numerically as the derivatives of the analytical first derivatives, and other times computes analytic frequencies. (See section 3.7 for details.) Whenever numerical second derivatives are computed after an SCF calculation-whether for frequency output, for an initial Hessian, or for updating during geometry optimization-the programs nude, onee, hfig, grid, rwr, scf, der1a, rwr, and der1b run, setting up and performing SCF calculations and evaluating analytic gradients at 6Natom perturbed geometries (unless the number of perturbed geometries needed is reduced by the use of molecular symmetry). To make each perturbed geometry, one atom is moved a small, fixed amount in the positive or negative direction along the x, y, or z Cartesian axes. After the calculations at the perturbed geometry, Jaguar performs one final calculation at the unperturbed geometry. (The Jaguar programs run may vary slightly for non-HF calculations, as described earlier in this section.) After the data from all perturbed geometries is collected, the program nude outputs the numerical first derivatives in a force table similar to the usual geometry optimization force table described earlier in this section. The output then lists the matrix indices of the most asymmetrical Hessian element before symmetrization. (The symmetrized numerical Hessian is not printed in the output, but can be found in the restart file, which is discussed in section 6.4.)
For either analytic or numerical frequency calculations, the output from the program freq contains the actual frequencies and normal modes from the computed Hessian, or from the last available Hessian (generally the initial Hessian guess) if you used the use available Hessian choice to request vibrational frequencies. The output from the program freq first lists the harmonic frequencies in cm1 and their symmetries (if symmetry is on for the job), then the normal modes. The system's thermochemical properties, the constant pressure heat capacity (Cp), entropy (S), enthalpy (H), and Gibbs free energy, are then listed for the specified pressure and temperatures, as well as at 0 K. Here is an example of this output from a vibrational frequency calculation on water:
If infrared intensities were calculated, several additional programs will run after the first run of the program scf. These programs compute the derivatives of the dipole, which are needed to calculate the IR intensities. The IR intensities themselves are listed in the frequencies table described above, which appears in the freq output near the end of the output file.
If your calculation uses a basis set that includes effective core potentials-that is, if the basis set's name begins with "LA"-the output lists the number of atoms treated with effective core potentials.
If the DIIS convergence method is not used, the "maximum DIIS error" column is not printed for the table giving data from the SCF iterations. Also, if the OCBSE convergence scheme is selected, the Coulomb and exchange contributions to the total two-electron terms are listed in the SCF summary beneath the table.
If a fully analytic calculations performed, as happens for some basis set selections (see section 3.8 for more details), the programs grid and rwr will not run, because the all-analytic method cannot take advantage of pseudospectral speedups.
If you select a Final localization method, the output from the program locpost appears after the output from any SCF iterations and lists the orbitals that are localized. (If you want to print out the localized orbitals, you should make the appropriate selection in the Orbitals window, as described in Section 5.7.)
The menu options from the Standard window, whose button appears under the Output heading, are described in this section. The output generated from these options will appear in the output file for the job. If you make a non-default setting from the Standard window, the output from the program pre will note the non-default options chosen. This output will appear above the molecular geometry output from the same program, and will indicate the non-default values of the keywords referred to in footnotes throughout this section.
If you turn this output option on, the output from the program pre will include an echo of the input file, a description of the path, which indicates the Jaguar programs run, and a list of keyword settings, including those made by default, and program parameters.1 This option is likely to be useful primarily for people who have a detailed knowledge of the code itself.
The memory information provided by this option is given for most of the routines used during the run, under the heading "dynamic memory statistics" each time.2 Current and maximum values for the number of arrays, their size in 8 byte words, and their size in bytes, as well as the type of variables used (e.g. real*8), are listed. The total and index i/o for the J and K matrices, in Mwords, are also provided after the energy output from all of the individual SCF iterations.
If you select this option, the cpu seconds spent in various Jaguar programs will be listed in the output.3
This option allows you to choose to print the geometry output in atomic units as well as in the usual units, Angstroms.4
If you choose to calculate multipole moments by making the appropriate setting in the Properties window, this option allows you to choose to list them in the output file in atomic units as well as in the usual units, Debye.5
When this option is turned on, the internuclear distances in Angstroms are listed for all nearest neighbor atoms in the output from the program pre, and the bond angles in degrees are given as well.6 The atoms are indicated with the atom labels assigned in the geometry input. When the Optimize geometry option in the Optimization window has been turned on, the bond lengths and angles standard output option is turned on automatically. For geometry optimizations, bond lengths and angles are also listed with the output from the program geopt.
The connectivity table provided by this option describes roughly how closely the atoms interact.7 Connectivity partially determines whether molecular fragments exist, the content of the initial Hessian, and many other properties of a calculation. The assignment of dealiasing functions for the pseudospectral method also depends upon the connectivities shown in this table, which reflect the neighbor ranges defined in the .daf file. (See section 9.3 for more information.) All of the diagonal entries are 0, indicating that the row atom and the column atom for the matrix element are the same atom. An entry of 1 indicates that the row atom and the column atom are considered to be bonded, because they are separated by a distance less than the sum of their covalent radii times the variable covfac, which is 1.2 by default and is also described in the Geometry Input Keywords subsection of section 8.6. If a connectivity table entry is 2, the corresponding row and column atoms are each bonded to some same third atom, by the definition of bonding described above. An entry of 3, 4, or more means that the atoms are within the third, fourth, or higher neighbor range of each other.
The overlap matrix S for the basis functions will be printed in five-column blocks if this option is selected.8 Since the matrix is symmetric, the elements within the top triangular half are not printed.
If the geometry optimization details9 option is selected, much additional information about the progress of a geometry optimization will be printed. This output often helps reveal the cause of any problems with optimizations.
The one-electron matrices representing kinetic energy and the sum of kinetic energy, nuclear attraction, and point charge-electron interactions will be printed in atomic orbital space in five-column blocks if this option is selected.10 Since the matrices are symmetric, the elements within the top triangular halves are not printed.
By turning this option on, you can choose to print out information about the Gaussian functions that make up the basis set.11 The orbitals in a basis set are made up of linear combinations of polynomials of the appropriate degree multiplied by Gaussian primitives of the form
, where N is a normalization constant and z is the exponent for the primitive. If the linear combination only includes one Gaussian primitive, the function is called uncontracted; otherwise, it is called a contracted Gaussian. The output controlled by this output option gives essentially the same information about the basis functions in two different tables, after giving a list of atoms and the basis set used for each one.
The shell information table is printed first. An example, for a calculation of water with a 631G** basis set, is given below. The first column of the table indicates which atom the primitive Gaussian being described in a particular row is centered upon. The second column lists the shell numbers, which increase consecutively for each atom. The values in the third column mean different things depending on their sign. The positive numbers mean that the basis function currently being described is composed of that number of primitive Gaussians, starting with the primitive Gaussian for that row and including the appropriate number of rows immediately beneath it. The negative numbers' magnitudes indicate the first shell which contributes to the same contracted Gaussian function. For instance, in the example below, the first row has a jcont value of 3, indicating that the first basis function being described is a contracted Gaussian composed of that primitive Gaussian and the two in the next two rows. The jcont values of 1 in the next two rows indicate that the primitive Gaussians being described are components in a contracted function whose first primitive Gaussian term is listed in the first row.
The values in the column marked "ishl" take on nonzero values when basis functions corresponding to different l values, as described in the next column, actually use primitive Gaussians with the same exponents. Positive values indicate that the same exponents should be used in the shell listed that number of rows down; a value of 1 indicates that the exponents should be provided from a shell listed earlier. The l values in the next column indicate angular momentum; a value of 1 corresponds to an s function, 2 indicates a p function, 3 a d function, and so on. The nfsh values are one less than the lowest number corresponding to the basis function or functions being described. Note that the nfsh=2 entries below are for p functions, so the fourth and fifth basis functions being described by the same rows are generated in the same way as the third (nfsh=2) one, except with different polynomials.
The column labeled z lists the exponents for the primitive Gaussians, while the "coef" column lists the coefficient of their contribution to the linear combination comprising the basis function. Note that the uncontracted basis functions, those with jcont values of 1, have "coef" values of exactly 1. Finally, the product of the "coef" value and the normalization constant for the primitive Gaussian, N, is listed in the column labeled "rcoef."
The second table, an example of which follows below, shows much the same information in a slightly different format. In this table, the different functions corresponding to an l value are listed explicitly-for instance, the entries X, Y, and Z for the seventh shell correspond to px, py, and pz orbitals. The only new information concerns the factors "rmfac," which may be needed to calculate each primitive Gaussian's contribution to the basis function. Sometimes, for l = 2 and higher, the value of "rcoef" calculated for the first primitive in the shell is different than it would have been if it had been calculated for a different primitive in that shell, and the "rmfac" values provide a way around that problem.
The table is followed by a list indicating the number of electrons in each atom which are treated with an effective core potential.
By turning this option on, you can choose to print out information about the derivatives of the basis set functions in terms of primitive Gaussians.12 The format and information is the same as that discussed for the Gaussian function list (basis set) option immediately above.
This section describes the menu options from the Files window, whose button appears under the Output heading. These output options do not alter the output file, but instead generate additional files. For each of the options described below, the relevant file will appear in the same directory as the output file. Each file's name is in the form jobname.suffix, where the different suffixes related to each kind of file are described below.
If you make a setting from the Files window, the output from the program pre will note the non-default options chosen. This output will appear above the molecular geometry output from the same program, and will indicate the non-default values of the keywords referred to in footnotes throughout this section.
When this option is selected, a file in the format of a Gaussian input file is created, with the suffix .g92.13 The file information includes the molecular geometry, the basis set name, and the type of calculation to be performed, as well as the molecular charge and the spin multiplicity of the molecule and any relevant effective core potential information. If symmetry is turned off, that setting will be entered into the .g92 file.
For GVB calculations, you should specify GVB pairs; Jaguar will also generate a GVB initial guess, which will be included in the .g92 file. For more information on setting up Gaussian input files, see section 6.6.
To write out an input file for the program GAMESS, you can select this option.14 The resultant file's suffix will be .gamess. The file will include the molecular geometry, the basis set, and some information on the type of calculation to be performed, as well as the molecular charge and the spin multiplicity of the molecule and any relevant effective core potential information.
You can use this option to generate a SPARTAN 4.0 archive file with the suffix .arc.15
If this option is turned on, a .gbs file will be generated containing the basis set in a form that can be used by Gaussian.16
If you set this option, Jaguar will create a file in XYZ format with the suffix .xyz.17 The file will contain all geometries generated during the course of the job (except that for solvated geometry optimizations, the file currently will only contain the solvated structures).
You can use this option to produce a file with the final orbitals in a format suitable for the program Molden [90].18
Some output can be printed out every SCF iteration by choosing options from the Per Iteration window, whose button appears under the Output heading. The output described in this section will appear in the output file. For each SCF iteration where the described output appears, that output is listed before the usual energy data for that iteration.
Any non-default settings from the Per Iteration window will cause the output from the program pre to note the non-default options chosen. This output will appear above the molecular geometry output from the same program, and will indicate the keywords referred to in footnotes throughout this section.
When this output option is off, the individual components contributing to the total energy are only printed for the final, converged result of the SCF iterations. When the option is turned on, the output includes each iteration's energy components: namely, the nuclear repulsion term, the total one-electron terms, the total two-electron terms, the electronic energy, and the total energy.19 The orbital energies for the occupied orbitals are also provided for each iteration.
The Coulomb and exchange contributions to the total two-electron energy will be printed as well if the J and K matrices are kept separate for the calculation, as for GVB calculations and when the Core J and K option in the Methods window is turned on. In addition, for most calculations involving open shells or higher-level methods, the individual contributions from each Hamiltonian are printed for the Coulomb and exchange terms.
If the calculation involves solvation, the nuclear-electronic and kinetic terms making up the one-electron terms will also be listed, as well as the term V/T (where V is the potential energy and T the kinetic energy) and the various contributions to the solvation energy.
If you select this option, the density matrix in atomic orbital space will be printed out for each iteration.20 For iterations in which Fock matrix updating is performed, using a matrix of elements calculated by taking the change in the density matrix from one iteration to the next, it is actually this differential density matrix which will be printed. The output from the program scf indicates whether Fock matrix updating was performed or not in any particular iteration.
The Coulomb and exchange matrices in atomic orbital space can be printed out for each iteration by selecting this option.21 However, by default the calculation will be performed by combining these matrices in the form 2J - K, and they may not be properly separated here if this is the case. In order to print out the true J and K matrices, you must insure that the Core J and K option in the Methods window, whose button is found in the main window, specifies that the matrices be kept separate. For GVB, DFT, LMP2, and GVB-LMP2 calculations, the J and K matrices are kept separate by default.
Since J and K are symmetric matrices, the elements from their top triangular halves are not printed.
The Fock matrix in atomic orbital space (for HF or DFT calculations) or molecular orbital space (for GVB calculations) can be printed by turning this option on.22 This information is only printed for iterations where the Fock matrix is not updated. Because the Fock matrix is symmetric, the elements from its top triangular half are not printed.
The Fock matrix in canonical orbital space can be printed by turning this option on.23 Because the Fock matrix is symmetric, the elements from its top triangular half are not printed.
You may print out GVB data for the initial guess and the GVB initial guess by selecting this option.24
Orbital information can be printed to the output file as well. Several possible choices are available in the Orbitals window, whose button is found under the Output heading, for what, when, and how orbitals should be printed.
If you choose to print out any orbital information, the output from the program pre will note the non-default options chosen. This output will appear above the molecular geometry output from the same program, and will indicate the keywords referred to in footnotes throughout this section.
To print out orbitals used for the HF initial guess, select this option.25
This option allows you to print out orbitals used for the GVB initial guess.26
The orbitals can be printed out after each SCF iteration in canonical orbital space.27 (Canonical orbital eigenvectors with very small eigenvalues are removed from the calculation before the SCF process.) The number of orbitals printed may depend upon the basis set, since some basis sets use five d functions for the canonical orbitals while others use six, as described in section 3.8.
The orbitals can be printed out after each SCF iteration in atomic orbital space.28
If this option is selected, orbitals can be printed in atomic orbital space after the SCF iterations.29
If Boys or Pipek-Mezey localization of the wavefunction has been requested using the button found in the main interface window, you can print the orbitals after the localization procedure by selecting this option.30
To print the orbitals at the end of the job, select this option.31
By default, no orbitals are printed in the output file, so the selection none will appear in the option menu labeled What.32 If you select occupied orbitals, all occupied orbitals, including GVB natural orbitals, will be printed.33 If the all orbitals option is selected, all occupied orbitals and ten virtual orbitals will be printed.34 (To change the default of ten virtual orbitals, see the information on the keyword ipvirt in section 8.6, under Orbital Output Keywords. The virtual orbitals are obtained by diagonalizing
, where f is the occupation of each orbital (1 for a closed shell).) Selection of GVB orbitals (nonorthog.) prints only the GVB non-orthogonal orbitals.35
The choices available for how to print the selected orbitals are large elements as f5.2, labels, in list,36 all elements as f10.5, labels, in table,37 all elements as f19.15, in list,38 all elements as f8.5, in list,39 and all elements as e15.6, in table.40 Examples of each of these style options appear below.
In the first option listed, the phrase "large elements" indicates that only coefficients larger than a particular value (generally .05) are listed. The notations "f5.2" and the like refer to standard FORTRAN formats. The word "labels" indicates that the atom identifiers (for instance, `h2') and the basis function types (for instance, S for s, Z for pz, or XX for dxx) are shown.
The output for each style is shown in either table form or list form. When the orbital output is in table form, each function's coefficient for each orbital is shown, with the functions shown in numbered rows and the orbitals in numbered columns. When it is in list form, each orbital is listed in turn, with the basis function coefficients listed in order. For the third and fourth options, those with f19.15 and f8.5 formatting, all coefficients are listed, in order but without numbering. The three styles presented in list form also include information on the occupation and energy of each orbital.
Because GVB orbitals are not computed until some time after the Hartree-Fock initial guess, you cannot choose to print GVB non-orthogonal orbitals if you have selected after HF initial guess above. Also, note that in canonical orbital space, the labels indicating atom identifiers and basis function types are meaningless.
If you generate output for occupied orbitals or all orbitals in the f19.15 or f8.5 formats, you can use it for input in the guess section of an input file, which is described in greater detail in section 8.11, or for input to Gaussian (guess=cards) or GVB2P5.
Here are some examples of output for each of these style options. The output shown is from output files generated from a calculation of water with a 631G** basis set, where the option requested under When was after SCF iterations and the option requested under What was occupied orbitals. Only the first two occupied orbitals are shown in each case, and not all functions are shown; these gaps are indicated by [...].
For the How option large elements as f5.2, labels, in list:
For the How option all elements as f10.5, labels, in table:
For the How option all elements as f19.15, in list:
For the How option all elements as f8.5, in list:
For the How option all elements as e15.6, in table:
The log file, an output file which appears in the local job directory, provides information on the progress of a run. You can display the current contents of a job's log file when you click the Check Job button found in the file viewer window, which can be opened by clicking Check from the main interface window. The log file notes when each program within Jaguar is complete, as well as noting data from each SCF iteration as it is calculated. The data from the SCF iterations is shown in table form. Some of the text for the column headings should be read down rather than across.
For the table of SCF iteration information, the number of the iteration is provided first in each row, followed by a "Y" or "N" indicating whether the Fock matrix was updated or not. When the Fock matrix is updated, the changes are made using a differential density matrix whose elements simply reflect the changes in the density matrix elements from the previous iteration to the current one.
The next entry indicates whether the DIIS convergence scheme was used for that iteration, also with a "Y" or "N". The DIIS method produces a new estimate of the Fock matrix as a linear combination of previous Fock matrices, including the one calculated during that iteration. DIIS, which is enabled by default, usually starts on the second iteration, and is not used on the final iteration. If the entry in this column reads "A", it indicates that DIIS was not used for that iteration, but the density matrix was averaged.
The cutoff set for each iteration is indicated under the "icut" heading. Cutoff sets are explained in the .cutoff file description in section 9.5.
The grid column lists the grid used for that iteration, which must be one of the grid types coarse (signified by a C), medium (M), fine (F), or ultrafine (U). See the subsection Grid and Dealiasing Function Keywords in section 8.6, and the description of the grid file in section 9.4, for more information on grids and grid types.
The total energy for the molecule in Hartrees appears in the next column, followed by the energy change, which is the difference in energy from the previous iteration to the current one.
The RMS density change column provides the root mean square of the change in density matrix elements from the previous iteration to the current one.
Finally, the maximum DIIS error column provides a measure of convergence by listing the maximum element of the DIIS error vector. For HF calculations, the DIIS error vector is given by FDS - SDF in atomic orbital space, where F, D, and S are the Fock, density, and overlap matrices, respectively. For open shell and GVB cases, the definition of the error vector is given in reference [11].
If you are not running a default, single-point, Hartree-Fock calculation, the log file will generally contain information generated from other Jaguar programs used for the run as well. This information is often a summary of what is written to the Jaguar output file. For a more detailed description of the information in the log file, you may wish to consult the previous sections of this chapter.
After all the individual programs necessary for that job have finished running, a note will appear in the log file listing the name and location of the output file. When the job is completed, this too is noted in the log file. At this point, if you wish, you can hit the View File button at the bottom of the file viewer window, selecting the appropriate output file name and hitting OK to see the output file for the job in the interface window.
33Relevant orbital output keyword set to 2, 3, 4, 5, or 6 in gen section of input file, depending on the format setting chosen.
34Relevant orbital output keyword set to 7, 8, 9, 10, or 11 in gen section of input file, depending on the format setting chosen.
35Relevant orbital output keyword set to 12, 13, 14, 15, or 16 in gen section of input file, depending on the format setting chosen.
36Relevant orbital output keyword set to 2, 7, or 12 in gen section of input file, depending on which orbitals are printed.
37Relevant orbital output keyword set to 3, 8, or 13 in gen section of input file, depending on which orbitals are printed.
38Relevant orbital output keyword set to 4, 9, or 14 in gen section of input file, depending on which orbitals are printed.
39Relevant orbital output keyword set to 5, 10, or 15 in gen section of input file, depending on which orbitals are printed.
40Relevant orbital output keyword set to 6, 11, or 16 in gen section of input file, depending on which orbitals are printed.
|
Schrödinger, Inc. http://www.schrodinger.com Voice: (503) 299-1150 Fax: (503) 299-4532 help@schrodinger.com Last updated: Thu Oct 11 19:10:29 2001 |
| |
|
|
|