Gaussian 98 Release Notes
Last Update: 10/28/2002

October 28, 2002

This document lists changes to and additional information about Gaussian 98 functionality since the printed documentation was prepared (the manual refers to the first printing of the Gaussian 98 User's Reference). The final section of these notes gives information about building and running Gaussian 98 on specific computer systems.

Section titles prefaced by an asterisk indicate items that have been updated or added for this release.

• With A.11.4, the Alpha Linux version is 64-bit.
• With A.11.3, there are changes in the Linux for Intel build procedure. See the System-Specific section for more information.
• Fujitsu PrimePower systems are now supported.
• There are now separate normal and Linda-enabled G98 binaries for Red Hat 6.2 and 7.2 (i.e., 4 versions). Similarly, there are also separate versions of the blas-f2c.a library for the two Red Hat versions. They are both included on the G98 source-code version distribution CD in the /intel top-level directory.
• Building the program on IA-64 systems requires the 6.0 beta version of the Intel compiler (build 20011129 as of this writing). The compiler can be obtained by sending email to joe.wolf@intel.com.

• The required citation has changed (see below).
• The 64-bit integer version is now the standard in the Hitachi and Hewlett-Packard environments. See the note on checkpoint file conversion in the system specific section of this document.
• The makefile has been updated to work with most recent Compaq Alpha compilers.
• Network parallel calculations may be performed on Compaq Alpha Linux systems via the Linda environment.
• Sorting routines now use full 64-bit addresses, so disk space limitations on 64-bit machines are removed.
• NEC systems no longer need to spawn separate subprocesses for each link. The NEC version now requires Super-UX Version 11.1.
• IBM Power 3 executables now use 64-bit pointers as well as integers. There are now four available IBM binary formats: Power 1 (32-bit), Power 2 (32-bit), Power 3 under AIX 4.3 (64-bit), and Power 3 under AIX 5.1 (64-bit). See the system-specific section for information on building in the various IBM environments, as this procedure has changed significantly. Note also that the correct compatible version of Linda must be used with each environment in order to run in parallel.

The required citation for Revision A.10 and higher is different from the one for all earlier Gaussian 98 revisions:

Gaussian 98 (Revision A.1x), M. J. Frisch, G. W. Trucks, H. B.Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A. Montgomery, Jr., R. E. Stratmann, J. C. Burant, S. Dapprich, J. M. Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi, V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K. Morokuma, P. Salvador, J. J. Dannenberg, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B. Foresman, J. Cioslowski, J. V. Ortiz, A. G. Baboul, B. B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin, D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, M. Challacombe, P. M. W. Gill, B. Johnson, W. Chen, M. W. Wong, J. L. Andres, C. Gonzalez, M. Head-Gordon, E. S. Replogle, and J. A. Pople, Gaussian, Inc., Pittsburgh PA, 2001.

Note: the x should be replaced by the corret integer for your version of Gaussian 98.

• The PBE1PBE (hybrid) and PBEPBE (pure) functionals have been added [29-30].
• MP2 frequencies may be computed in the presence of an external electric field (i.e., MP2 Freq Field=…).
• The Freq=Numer Polar Field=… keyword combination is now valid.
• The IRC(StepSize=n) now sets the convergence criteria correctly instead of making them too loose.
• By default, SCI-PCM now turns off charge compensation (as was the case in Gaussian 94).
• The selection of internal coordinates has been improved for optimizations, resulting in better behavior for systems like weakly-bound complexes.
• The printing of transition densities between excited states using the Time Dependant DFT and HF methods is disabled because not all terms are currently computed. The ground-to-excited state transition densities are correct and are still printed.
• The incorrect exponent for N in the D95 basis has been corrected (results in a 10 microhartree energy change for the atom).
• The special atom types O_3_z, P_3+5, P_3+q, S_3+4, S_3+6, and W_3+6 in the Dreding/UFF force fields are now given the correct values. Previously, they were given the default parameters (e.g. O_3's for O_3_z).
• The PM3 parameter for Li has been updated per reference [31].

NBO is now dimensioned for 200 atoms and 10000 basis functions.

• The TD DFT procedures for pure DFT now does the symmetric problem (a faster algorithm)
• The initial guess for TD DFT does not use improved virtual orbitals (IVOs).
• Using TD=Read to read an initial guess from a calculation on using a pure functional may now be use for a subsequent hybrid DFT functional calculation.
• CIS(Add=n) and TD(Add=n) now imply the Read option as well (they were meaningless without it).

Counterpoise corrections may be computed using the Counterpoise keyword (which can be used on an energy calculation, optimization or frequency calculation). The facility cannot be used with ECPs or a general basis set.

The Counterpoise keyword takes an integer value specifying the number of fragments or monomers in the molecular structure. The facility also requires an additional integer to be placed at the end of each atom specification indicating which fragment/monomer it is part of.

Here are examples using a Z-matrix (left) and Cartesian coordinates (right):

# MP2/6-31G Counterpoise=2 Opt              # MP2/6-31G Counterpoise=2 Opt

Counterpoise with Z-matrix                  Counterpoise with Cartesians

0,1,0,3,1,2                                 0,1
O,0.0,0.0,0.0,1                             1  0.00  0.00  0.92  1
O,1,ROO,2                                   9  0.17  0.00  2.73  2
X,1,1.,2,X3O                                1  0.77  0.00  3.43  2
H,1,RO1H,3,HOX3,2,90.,0,1                   9  0.00  0.00  0.00  1
H,1,RO1H,3,HOX3,2,-90.,0,1
X,2,1.,1,52.5,3,180.,0
H,2,RO2H1,6,H7OX,1,180.,0,2
H,2,RO2H2,6,H8OX,1,0.,0,2

Note that the Z-matrix input requires a 0 after the dihedral angle value/variable (to indicate that the final angle is a dihedral) prior to the fragment number. Also, the first atom in the Z-matrix must be given in Cartesian coordinates. Clearly, using Cartesian coordinates for such jobs makes specifying fragment numbers in the input much more straightforward.

The preceding Z-matrix also illustrates the use of fragment-specific charge and spin multiplicity specifications. The format of the corresponding input line in this case is:

molecule-charge,  molecule-spin multiplicity,f  rag. 1-charge,  frag.1 multiplicity,  frag. 2 charge,  frag. 2 multiplicity

G98 Revision A.10 can carry out a PCM calculation using Klamt's form of the conductor reaction field (COSMO) and generate the input data for the COSMO-RS solubility programs.

The SCRF=COSMORS keyword requests a conductor PCM calculation (CPCM) using atomic radii and other parameters as suggested by Klamt for his models. The name of the text file to write with input data for COSMO-RS is read from the input stream, after the geometry, basis set and other data.

Only single-point calculations are possible with COSMORS option. These calculations will typically be done as single-point solvated calculations using SCRF=CPCM optimized geometries.

Here is a sample input file:

# B3LYP/6-311+G(2d,2p) SCF=(Tight) SCRF=COSMORS
 
Water generating COSMO-RS input

0 1
o
h,1,r
h,1,r,2,a

r .96
a 104.5

water.cosmo

This produces the data file water.cosmo.

COSMO-RS is distributed as COSMOtherm by COSMOlogic GmbH, www.cosmologic.de.

This table from the introduction to chapter 3 of the manual has been updated. The new table is printed below:

Gaussian 98 Input Section Ordering

*A blank line also separates the second or third title section from the corresponding molecule specification.
**UHF jobs use separate a andb sections (themselves separated by a blank line).

The reported number of primitive gaussians will differ between revisions A.4 and higher and previous revisions. Note that this change has no effect on the results of the calculation since it is an arbitrary, completely informational statistic not used in any actual computation. Previously, this value indicated the number of AO components of gaussian primitives in terms of pure d functions if you were using pure contracted functions in the calculation. Now, it always counts Cartesians regardless of the types of contracted functions in use.

The default memory amount in G98 is 6MW.

The EPR-II and EPR-III basis sets of Barone [1] are included in Gaussian 98. They are optimized for the computation of hyperfine coupling constants by DFT methods (particularly B3LYP). EPR-II is a double zeta basis set with a single set of polarization functions and an enhanced s part: (6,1)/[4,1] for H and (10,5,1)/[6,2,1] for B to F. EPR-III is a triple-zeta basis set including diffuse functions, double d-polarizations and a single set of f-polarization functions. Also in this case the s-part is improved to better describe the nuclear region: (6,2)/[4,2] for H and (11,7,2,1)/[7,4,2,1] for B to F. The basis sets are available for H, B, C, N, O and F.

The following options to select the numerical integration weighting scheme have been added to the Integral keyword:

SSWeights
Use the weighting scheme of Scuseria and Stratman [3] for the numerical integration for DFT calculations. This is the default.

BWeights
Use the weighting scheme of Becke for numerical integration. This was the default in Gaussian 94.

The description of the IRCMax keyword has been updated:

Description

Performs an IRCMax calculation using the methods of Petersson and coworkers [4-12]. Taking a transition structure as its input, this calculation type finds the maximum energy along a specified reaction path.

Required Input

IRCMax requires two model chemistries as its options, separated by a colon:IRCMax(model2: model1). Here is an example route section:

# IRCMax(B3LYP/6-31G(d,p):HF/6-31G(d,p))

This calculation will find the point on the HF/6-31G(d,p) reaction path where the B3LYP/6-31G(d,p) energy is at its maximum.

The Zero option will produce the data required for zero curvature variational transition state theory zero curvature variational transition state theory (ZC-VTST) [5, 6, 9-12]. Consider the following route:

# IRCMax(MP2/6-31G(d):HF/3-21G(d),Zero,Stepsize=10)

This job will start from the HF/3-21G(d) TS and search along the HF/3-21G(d) IRC with a stepsize of 0.1 amu-1/2 bohr until the maximum of the MP2/6-31G(d) energy (including the HF/3-21G(d) ZPE) is bracketed. The position along the HF/3-21G(d) IRC for this MP2/6-31G(d) TS will then be optimized. The output includes all quantities requred for the calculation of reaction rates using the ZC-VTST version of absolute rate theory: TS moments of inertia, all real vibrational frequencies (HF/3-21G(d)), the imaginary frequency for tunneling (fit to MP2/6-31G(d) + ZPE), and the total MP2/6-31G(d) + ZPE energy of the TS.

ZC-VTST Options

Zero
Include the zero-point energy in the IRCMax computation.

Path Selection Options

Forward
Follow the path only in the forward direction.

Reverse
Follow the path only in the reverse direction.

ReadVector
Read in the vector to follow. The format is Z-matrix (FFF(I), I=1,NVAR), read as (8F10.6).

MaxPoints option=N
Number of points along the reaction path to examine (in each direction if both are being considered). The default is 6.

StepSize option=N
Step size along the reaction path, in units of 0.01 amu-1/2-Bohr. The default is 10.

MaxCyc option=N
Sets the maximum number of steps in each geometry optimization. The default is 20.

Freq
Calculate the projected vibrational frequencies for motion perpendicular to the path, for each optimized point on the path [13]. This option is valid only for reaction paths in mass-weighted internal coordinates.

Coordinate System Selection Options

MassWeighted
Follow the path in mass-weighted internal (Z-matrix) coordinates (which is equivalent to following the path in mass-weighted Cartesian coordinates)mass-weighted internal coordinates. MW is a synonym for MassWeighted. This is the default.

Internal
Follow the path in internal (Z-matrix) coordinates without mass weighting.

Cartesian
Follow the path in Cartesian coordinates without mass weighting.

Isotope Specification Option

ReadIsotopes
Specify alternate isotopesisotopes (the defaults are the most abundant isotopes). This information appears in a separate input section having the format:

isotope mass for atom 1
isotope mass for atom 2
...
isotope mass for atom N

where the lines hold the isotope masses for the various atoms in the molecule, arranged in the same order as they appeared in the molecule specification section. If integers are used to specify the atomic masses, the program will automatically use the corresponding actual exact mass (e.g., 18 specifies O18, and Gaussian 98 uses the value 17.99916).

Convergence-Related Option

VeryTight
Tightens the convergenceIRC criteria used in the optimization at each point along the path. This option is necessary if a very small step size along the path is requested.

Options for Generating Initial Force Constants

CalcFC
Specifies that the force constants be computed at the first point

CalcAll
Specifies that the force constants be computed at every point.

FCCards
Reads the Cartesian forces and force constants from the input stream after the molecule specifications. This option can be used to read force constants recovered from the Quantum Chemistry Archive using its internal FCList command. The format for this input is:

Energy (format D24.16)
Cartesian forces (lines of format 6F12.8)
Force constants (lines of format 6F12.8)

The force constants are in lower triangular form: ((F(J,I),J=1,I),I=1,NAt3), where NAt3 is the number of Cartesian coordinates. If both FCCards and ReadIsotopes are specified, the masses of the atoms are input before the energy, Cartesian gradients and the Cartesian force constants.

Note that the RCFC option is not supported with IRCMax.

Restart Option

Restart
Restarts an IRC calculation which did not complete, or restarts an IRC calculation which did complete, but for which additional points along the path are desired.

Availability

Analytic gradients are required for the IRC portion of the calculation (model1 above). Any non-compound energy method and basis set may be used for model2.

Related Keywords

IRC, Opt, Freq

MP3, MP4, QCISD, CCSD, QCISD(T), and CCSD(T) calculations all now look at Maxdisk. If the calculation can be done using a full integral transformation while keeping disk usage under MaxDisk, this is done; if not, a partial transformation is done and some terms are computed in the AO basis. Since MP2 obeys MaxDisk as much as possible, the Stingy, NoStingy and VeryStingy options are not needed.

Thus, it is crucial for a value for MaxDisk to be specified explicitly for these types of jobs, either within the route section or via a system wide setting in the Default.Route file. If MaxDisk is left unset, the program now assumes that disk is abundant and performs a full transformation by default, in contrast to G94 where a partial transformation was the default in such cases. If MaxDisk is not set and sufficient disk space is not available for a full transformation, the job will fail (where it may have worked in G94).

The Polar keyword has an additional option:

Analytic
Compute polarizability and hyperpolarizability analytically. This is possible for RHF and UHF for which it is the default. The polarizability is always computed during analytic frequency calculations.

The Prop keyword now supports the EPR option:

EPR
Compute the anisotropic hyperfine coupling constants [1, 14, 15].

The PCM models are available for HF, DFT, MP2, MP3, MP4(SDQ), QCISD, CCD, CID, and CISD energies and HF and DFT gradients.

References for the ZINDO method are now provided [16-24].

The norm of the density gradient and the Laplacian are also scalar (i.e., one value per point), and are written out in the same manner. Density+gradient grids are similar, but with two writes for each row (of lengths N3 and 3*N3).

Note that all cube files are formatted; requests for unformatted cube files are currently ignored.

Trajectory calculations may be performed at the MP2 level as well as the methods listed in the manual.

The use of microiterations in geometry optimizations is the default for MM optimizations and ONIOM optimizations with an MM component. Use the Opt=NoMicro option to turn off microiterations.

The command formchk -c causes the molecular mechanics atom types to appear in the formatted checkpoint file as strings rather than integers.

The SDD basis set keyword consists of D95 up to Ar and then a variety of Stuttgart potentials for Z > 18 (see the manual). The SDDAll keyword selects Stuttgart potentials for Z > 2.

The SDD, SHF, SDF, MHF, MDF, MWB forms may be used to specify these basis sets/potentials within Gen basis input. Note that the number of core electrons must be specified.

The Midi! basis set (keyword MidiX) is now implemented for H, C-F, Si-Cl, I and Br.

The CBS-QB3 keyword may be used to select the CBS-QB3 method [25]. The CBS-QB3 method has been updated to use the new and more stable Minimal Population localization procedure. This will cause small differences in the CBS-QB3 energy computed with pre-A.7 versions of Gaussian 98, but it is more reliable for large molecules. The CBS-QB3O keyword requests the CBS-QB3 model with the earlier localization scheme, if comparison with previous results is necessary.

The CBS-4 model chemistry has been updated with both the new localization procedure (discussed in the preceding item) and improved empirical parameters. The new version CBS-4M (M referring to the use of Minimal Population localization) is recommended for new studies; the CBS-4O keyword requests the earlier parametrization. Since these can give significantly different results the previous CBS-4 keyword prints an error message rather than defaulting to either version.

The Geom=PrintInputOrient option has been added to include the table giving the cartesian coordinates in the input orientation. By default, this table is printed for smaller molecules but omitted for large ones.

The NMR=PrintEigenvectors option has been added to display the eigenvectors of the shielding tensor for each atom.

The recently published G3 [26] and G3MP2 [27] methods have been added, along with the variants using B3LYP structures and frequencies, G3B3and G3MP2B3 [28] (using these keywords).

The generation of redundant internal coordinates for weakly bound complexes has been updated. The "HBond" and "AllHBond" options never did what the documentation suggested (and they have been removed); what is done in revision A.7 is to include Hydrogen bonds automatically. In addition, in connecting different fragments which are only weakly bound (hydrogen-bonded and otherwise), all pairs of atoms with one atom in each fragment having distance within a factor of 1.3 of the closest pair have their distances added to the internal coordinates. If at least 3 such pairs are found, then no angles or dihedrals involving both fragments are added. However, if only 1 or two pairs of atoms are close, then the related angles and dihedrals are added in order to ensure a complete coordinate system. As usual, the ModRedundant option can be used to add or remove any coordinates manually.

The default generation of charges when using the UFF force field was inconsistent and caused considerable confusion. In Revision A.7, no charges are assigned to atoms by default when using any molecular mechanics force field. Options are available to estimate charges at the initial point using the QEq algorithm under control of the following options for any of the mechanics keywords:

Link 0 commands (% lines) can now be up to 500 characters long.

• The IRC keyword supports a CalcAll option.
• The only valid separator between the various methods for an IRCMax calculation is a colon. In particular, a double slash (//) will not work.
• The Density is ignored when used with ZINDO.
• Defaults for the FMM keyword have been set.
• The default algorithm for optimizations without gradients is EF.
• Charge obeys the Units keyword, contrary to manual.
• The example job in the ONIOM discussion use the Amber MM method without specifying atom types in the molecule specification. This will generate an error since Amber always requires that explicit atom types be input. Changing Amber to UFF in this job will make the example run correctly.
• The DFT keyword "G961LYP" is not implemented.
• The keyword sequence "MP2 NMR=GIAO" is broken and performs only an MP2 energy calculation. Use MP2 NMR to perform a GIAO NMR calculation at the MP2 level.
• Int=NoSComp may be used to turn off symmetry blocking of MO 2-electron integrals. NoSymmComp is a synonym for NoSComp.
• The BiMolecularReactant and BiMolecularProduct options to the Opt keyword are not implemented. The BiMolecular option is provided, however.
• Opt(StepSize=n) does not work. One must use IOp(1/8=n) instead.

A few features documented in the manual will not appear in Gaussian 98:

• The Becke96 gradient-corrected exchange functional (accordingly, B1B96 is also deferred).
• The B1PW91 hybrid functional.
• Support for compound energy methods with IRCMax.
• Support for the Phase, Freq and Downhill options to IRC.

The following reference citations have been updated/corrected:

123 C. Peng, P. Y. Ayala, H. B. Schlegel and M. J. Frisch, "Using redundant internal coordinates to optimize geometries and transition states," J. Comp. Chem. 17, 49 (1996)

132 S. Dapprich, I. Komaromi, K. S. Byun, K. Morokuma and M. J. Frisch, "A New ONIOM Implementation in Gaussian 98. Part I. The Calculation of Energies, Gradients, Vibrational Frequencies and Electric Field Derivatives," J. Mol. Str. (Theochem) 461-462, 1 (1999). [replaces manual references 132 and 133]

146 R. E. Stratmann, J. C. Burant, G. E. Scuseria and M. J. Frisch, J. Chem. Phys. 106, 10175 (1997).

319 C. Adamo and V. Barone, Chem. Phys. Lett. 274, 242 (1997).

1 V. Barone, in Recent Advances in Density Functional Methods, Part I, Ed. D. P. Chong (World Scientific Publ. Co., Singapore, 1996).

3 E. Stratmann, G. E. Scuseria and M. J. Frisch, Chem. Phys. Lett. 257, 213 (1996).

4 D. K. Malick, G. A. Petersson and J. A. Montgomery Jr., "Transition States for Chemical Reactions. I. Geometry and Barrier Height," J. Chem. Phys. 108, 5704 (1998).

5 B. C. Garrett, D. G. Truhlar, R. S. Grev and A. D. Magnusson, "Improved treatment of threshold contributions in variational transition state theory," J. Phys. Chem. 84 (1980).

6 G. A. Petersson, "Complete Basis Set Thermochemistry and Kinetics," in Computational Thermochemistry, Ed. K. K. Irikura and D. J. Frurip (Amer. Chem. Soc., Washington, DC, 1998) 237.

7 M. Schwartz, P. Marshall, R. J. Berry, C. J. Ehlers and G. A. Petersson, "Computational Study of the Kinetics of Hydrogen Abstraction from Fluoromethanes by the Hydroxyl Radical," J. Phys. Chem. , submitted (1998).

8 G. A. Petersson, D. K. Malick, W. G. Wilson, J. W. Ochterski, J. A. Montgomery Jr. and M. J. Frisch, "Calibration and comparison of the G2, CBS, and DFT methods for computational thermochemistry," J. Chem. Phys., 109, 10570 (1998).

9 H. Eyring, "The activated complex in chemical reactions," J. Chem Phys. 3, 107 (1935).

10 D. G. Truhlar, "Adiabatic Theory of Chemical Reactions," J. Chem. Phys. 53, 2041 (1970).

11 D. G. Truhlar and A. Kuppermann, "Exact tunneling calculations," J. Am. Chem. Soc. 93, 1840 (1971).

12 R. T. Skodje, D. G. Truhlar and B. C. Garrett, "Vibrationally adiabatic models for reactive tunneling,"J. Chem. Phys. 77, 5955 (1982).

13 A. G. Baboul and H. B. Schlegel, "Improved Method for Calculating Projected Frequencies along a Reaction Path," J. Chem. Phys. (1997).

14 N. Rega, M. Cossi and V. Barone, J. Chem. Phys. 105, 11060 (1996).

15 V. Barone, Chem. Phys. Lett. 262, 201 (1996).

16 A. D. Bacon and M. C. Zerner, "An Intermediate Neglect of Differential Overlap Theory for Transition Metal Complexes: Fe, Co, and Cu Chlorides," Theo. Chim. Acta 53, 21 (1979).

17 W. P. Anderson, W. D. Edwards and M. C. Zerner, "Calculated Spectra of Hydrated Ions of the First Transition-Metal Series," Inorganic Chem. 25, 2728 (1986).

18 M. C. Zerner, G. H. Lowe, R. F. Kirchner and U. T. Mueller-Westerhoff, "An Intermediate Neglect of Differential Overlap Technique for Spectroscopy of Transition-Metal Complexes. Ferrocene," J. Am. Chem. Soc. 102, 589 (1980).

19 J. E. Ridley and M. C. Zerner, "An Intermediate Neglect of Differential Overlap Technique for Spectroscopy: Pyrrole and the Azines.," Theo. Chim. Acta. 32, 111 (1973).

20 J. E. Ridley and M. C. Zerner, "Triplet states via Intermediate Neglect of Differential Overlap: Benzene, Pyridine, and the Diazines," Theo. Chim. Acta. 42, 223 (1976).

21 M. A. Thompson and M. C. Zerner, "The Electronic Structure and Spectroscopy of the Photosynthetic Reaction Center from Rhodopseudomonas viridis," J. Am. Chem. Soc. 113, 8210 (1991).

22 M. C. Zerner, "Semi Empirical Molecular Orbital Methods," in Reviews of Computational Chemistryd, Ed. K. B. Lipkowitz and D. B. Boyd (VCH Publishing, New York, 1991), vol. 2, 313.

23 M. C. Zerner, P. Correa de Mello and M. Hehenberger, "On the Convergence of the Self Consistent Field Method to Excited States," Int. J. Quant. Chem. 21, 251 (1982).

24 L. K. Hanson, J. Fajer, M. A. Thompson and M. C. Zerner, "Enviromental Effects on the Properties of Bacteriachlorphylls in Photosynthetic Reaction Centers: Theoretical Models," J. Amer. Chem. Soc. 109, 4728 (1987).

25 J. A. Montgomery Jr, M. J. Frisch, J. W. Ochterski and G. A. Petersson, "A complete basis set model chemistry. VI. Use of density functional geometries and frequencies.," J. Chem. Phys. 110, 2822 (1999).

26 L.A. Curtiss, K. Raghavachari, P.C. Redfern, V. Rassolov and J.A. Pople, J. Chem. Phys., 109, 7764 (1998).

27 L.A. Curtiss, P.C. Redfern, K. Raghavachari, V. Rassolov and J.A. Pople, J. Chem. Phys., 110, 4703 (1999).

28 A.G. Baboul, L.A. Curtiss, P.C. Redfern, and K. Raghavachari, J. Chem. Phys., 110, 7650 (1999).

29 J.P.Perdew,K.Burke and M.Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.

30 J.P.Perdew,K.Burke and M.Ernzerhof, Phys. Rev. Lett. 78 (1997) 13965.

31 E. Anders, R. Koch, and P. Freunscht, J. Comp. Chem. 14 (1993) 1301.

System-Specific Information

Be sure to check the Gaussian web site regularly for information about requred operating system and compiler revision levels for systems upon which you want to build the program. Any patches required by specific hardware and/or operating system levels are also listed here.

You can go directly to this page with the URL: www.gaussian.com/g98_req.htm.

Sun users should be using the cc command from /opt/SUNWspro/bin and not from /usr/ucb. Be sure that the first directory is included in your path and that it appears in it before /usr/ucb.

By default, the SGI R10000, Sun and AIX Power 3 versions are built using 64-bit integers. This removes the previous limits of 16 GB of main memory and 16 GB disk space imposed in previous versions, but means that checkpoint files from earlier revisions of Gaussian 98 cannot be used directly by Revision A.8 or higher in the case of SGI, Sun, Compaq Tru-64 and IBM or by Revision A.10 in the case of Hitachi and Hewlett-Packard.

To convert old IBM, Sun or SGI checkpoint files, use the copy of formchk or chkmove from revision A.6 to convert them to formatted files, then use unfchk or chkmove from A.8 to convert them back to binary form. In the case of Hitachi, use the copy of formchk or chkmove from revision A.9 or earlier to convert old binary checkpoint files to formatted files, then use unfchk or chkmove from A.10 to convert them back to binary form .

The procedure for compiling Gaussian 98 from source code on IBM RS/6000 systems has been changed for A.10. Because of difficulties with IBM's ESSL library, we have switched to the public-domain ATLAS libraries for turned matrix operation subroutines. Since building ATLAS can be cumbersome, we have included the ATLAS libraries on the G98 source CD, in the /atlas top-level directory.

The G98 makefile expects to find the two ATLAS libraries in /usr/local/lib. Thus, before compiling G98, the appropriate versions of the libraries should be copied to this location. If a different location is chosen for these libraries, then the file g98/bsd/rs6k.make must be edited to reflect the change. The libraries provided are:

power1-4.3-libatlas.a
power1-4.3-libf77blas.a
power2-4.3-libatlas.a
power2-4.3-libf77blas.a
power3-4.3-libatlas.a
power3-4.3-libf77blas.a
power3-5.1-libatlas.a
power3-5.1-libf77blas.a

Note that G98 is built as a 64-bit program for the Power 3 machines, and that AIX 4.3 and AIX 5.1 have different and incompatible binary formats for 64-bit programs. Consequently, there are four pairs of libraries:

For example, for a Power 3 machine running AIX 4.3, execute the following commands (the percent sign is the C shell prompt and should not be typed):

  % cp /CDROM/atlas/power3-4.3-libf77blas.a /usr/local/lib/libf77blas.a
  % cp /CDROM/atlas/power3-4.3-libatlas.a /usr/local/lib/libatlas.a

where /CDROM should be replaced with the path to the CDROM drive.

Once the proper libraries are copied, the program can be built as follows:

Power 3 systems:
  % cd $g98root/g98
  % source bsd/g98.login
  % bsd/bldg98 >& g98_build.log

Power 2 systems:
  % cd $g98root/g98
  % source bsd/g98.login
  % bsd/bldg98 all ibmp2 >& g98_build.log 

Power 1 systems:
  % cd $g98root/g98
  % source bsd/g98.login
  % bsd/bldg98 all ibmp1 >& g98_build.log 

Process and job memory limits should be set to 0; otherwise, the default memory used will be 90% of all available memory (rather than 4MB as in other versions of Gaussian 98). This can be done by giving input similar to the following to udbgen:

update:username:pmemlim[b]:0:jmemlim[b]:0:jmemlim[i]:0:

G98 takes more fixed memory on UniCOS systems. When running in NQS batch queues, make sure that the NQS batch memory limit is at least 6 million words larger than the value desired or specified for %Mem.

The Linux version requires the Portland Group Fortran compiler (see www.pgroup.com).

The procedure for compiling Gaussian 98 from source code on Intel Linux systems has been changed for A.11.3. We use the public-domain ATLAS libraries for turned matrix operation subroutines. Since building ATLAS can be cumbersome, we have included the ATLAS libraries on the G98 source CD, in the /atlas top-level directory.

The G98 makefile expects to find the two ATLAS libraries in /usr/local/lib. Thus, before compiling G98, the appropriate versions of the libraries should be copied to this location. If a different location is chosen for these libraries, then the file g98/bsd/i386.make must be edited to reflect the change. The libraries provided are PIII-RH7-libatlas.a and PIII-RH7-libf77blas.a.

These libraries can be used with any Linux version (we've tested with Red Hat 6 as well), and with the Pentium 4 as well.

You can install the libraries from the CD using the following commands (the percent sign is the C shell prompt and should not be typed):

   % cp /CDROM/atlas/PIII-RH7-libf77blas.a /usr/local/lib/libf77blas.a
   % cp /CDROM/atlas/PIII-RH7-libatlas.a /usr/local/lib/libatlas.a

where /CDROM should be replaced with the path to the CDROM drive.

Once the proper libraries are copied, the program can be built as follows:

  % cd $g98root/g98
  % source bsd/g98.login
  % bsd/bldg98 >& g98_build.log

Do not set the Windows environment variable MKL_NPROCS to a value greater than 1 in an attempt to make the program run in parallel. This will result in the program still using only a single processor, but a significant amount of gratuitous overhead will be expended anyway.

The Linux version of the program for the Compaq Alpha platform must be built using the Compaq Fortran compiler. GNU compilers will not work.

By default, the Fujitsu version of the Gaussian 98 makefile ($g98root/g98/bsd/fujitsu.make) is configured to build the VPP800/VPP5000 version of the code. VX/VPP300/VPP700/VX users must edit this file according to the instructions contained within it.