Manual:Basis:Large component basis set

From Diracwiki

Jump to: navigation, search

Contents

Large component basis set

The specification of the large component basis set follows the atomic coordinates. There are various possibilities for giving the basis set.

Basis sets from the basis set library 

DIRAC
Water
aug-cc-pVDZ basis (note that a tight p has NOT been added)
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE BASIS aug-cc-pVDZ

If you use a aug-cc-p... basis set, you can prefix it with d-, t-, q-, and it will add the diffuse functions in an even-tempered fashion.

Multiple basis sets from the basis set library

As an alternative to the BASIS option described above, it is possible to use specify different basis set files via the MULTIBASIS keyword. The MULTIBASIS option is present to allow easier inclusion of different sets of diffuse and/or polarization functions to a reference basis set. The syntax for this keyword is very similar to that of the BASIS: 

DIRAC
Water
aug-cc-pVDZ basis (note that a tight p has NOT been added)
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE MULTIBASIS 2 cc-pVDZ cc-pVDZ-diffuse

After the keyword an integer with the number of files to be read should be specified. It is followed by the name of the different basis set files, each separated by a whitespace. The only limitation for the number of basis set files is that the total lenght of this line should not exceed Fortran's maximum allowed line size (80 characters).

Explicitly typed basis sets

Explicitly typed basis sets are best described using an explicit example: 

DIRAC
Water
aug-cc-pVDZ basis (note that a tight p has NOT been added)
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EXPLICIT  3    1    1    1
# s functions
f  10    4
  11720.0000000  0.00071000 -0.00016000  0.00000000  0.00000000
   1759.0000000  0.00547000 -0.00126300  0.00000000  0.00000000
    400.8000000  0.02783700 -0.00626700  0.00000000  0.00000000
    113.7000000  0.10480000 -0.02571600  0.00000000  0.00000000
     37.0300000  0.28306200 -0.07092400  0.00000000  0.00000000
     13.2700000  0.44871900 -0.16541100  0.00000000  0.00000000
      5.0250000  0.27095200 -0.11695500  0.00000000  0.00000000
      1.0130000  0.01545800  0.55736800  0.00000000  0.00000000
      0.3023000 -0.00258500  0.57275900  1.00000000  0.00000000
      0.0789600  0.00000000  0.00000000  0.00000000  1.00000000
# p functions
f   5    3
     17.7000000  0.04301800  0.00000000  0.00000000
      3.8540000  0.22891300  0.00000000  0.00000000
      1.0460000  0.50872800  0.00000000  0.00000000
      0.2753000  0.46053100  1.00000000  0.00000000
      0.0685600  0.00000000  0.00000000  1.00000000
# d functions
f   2    2
      1.1850000  1.00000000  0.00000000
      0.3320000  0.00000000  1.00000000

Highest angular quantum number l plus one. In this case it is 3, since we are using a spd basis set.

Number of blocks for each l-value. The memory requirements grow rapidly with the number of basis functions in a block (note for instance that four g functions actually are 60 basis functions, as there are 15 cartesian components of each g function). Memory requirements can therefore be reduced by splitting basis functions of the quantum number into different blocks. This will, however, decrease the performance of the integral calculation.

Lines starting with either !, $, or # are comments.

Number of primitive Gaussians in this block.

Number of contracted Gaussians in this block. If a zero is given, an uncontracted basis set will be assumed, and only orbital exponents need to be given. 

DIRAC
Water
aug-cc-pVDZ basis (note that a tight p has NOT been added)
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EXPLICIT  3    1    1    1
# s functions
f  10    4
  11720.0000000  0.00071000 -0.00016000  0.00000000  0.00000000
   1759.0000000  0.00547000 -0.00126300  0.00000000  0.00000000
    400.8000000  0.02783700 -0.00626700  0.00000000  0.00000000
    113.7000000  0.10480000 -0.02571600  0.00000000  0.00000000
     37.0300000  0.28306200 -0.07092400  0.00000000  0.00000000
     13.2700000  0.44871900 -0.16541100  0.00000000  0.00000000
      5.0250000  0.27095200 -0.11695500  0.00000000  0.00000000
      1.0130000  0.01545800  0.55736800  0.00000000  0.00000000
      0.3023000 -0.00258500  0.57275900  1.00000000  0.00000000
      0.0789600  0.00000000  0.00000000  0.00000000  1.00000000
# p functions
f   5    3
     17.7000000  0.04301800  0.00000000  0.00000000
      3.8540000  0.22891300  0.00000000  0.00000000
      1.0460000  0.50872800  0.00000000  0.00000000
      0.2753000  0.46053100  1.00000000  0.00000000
      0.0685600  0.00000000  0.00000000  1.00000000
# d functions
f   2    2
      1.1850000  1.00000000  0.00000000
      0.3320000  0.00000000  1.00000000

A single character describing the input format of the basis set in this block.

The default format is 8F10.4 which will be used if left blank. Be very careful when using this default format as it will miss any exponential parameter standing to the right of the 10 characters. In this format the first column is the orbital exponent and the seven last columns are contraction coefficients. If no numbers are given, a zero is assumed. If more than 7 contracted functions occur in a given block, the contraction coefficients may be continued on the next line, but the first column (where the orbital exponents are given) must then be left blank.

An F or f in the first position (like in the example above) will indicate that the input is in free format. This will of course require that all contraction coefficients need to be typed in, as all numbers need to be present on each line. However, note that this options is particularly handy together with completely decontracted basis sets, as described below. Note that the program reads the free format input from an internal file that is 80 characters long, and no line should therefore exceed 80 characters.

One may also give the format H or h. This corresponds to high precision format 4F20.8, where the first column again is reserved for the orbital exponents, and the three next columns are designated to the contraction coefficients. If no number is given, a zero is assumed. If there are more than three contracted orbitals in a given block, the contraction coefficients may be continued on the next line, though keeping the column of the orbital exponents blank. 

DIRAC
Water
aug-cc-pVDZ basis (note that a tight p has NOT been added)
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EXPLICIT  3    1    1    1
# s functions
f  10    4    0
  11720.0000000  0.00071000 -0.00016000  0.00000000  0.00000000
   1759.0000000  0.00547000 -0.00126300  0.00000000  0.00000000
    400.8000000  0.02783700 -0.00626700  0.00000000  0.00000000
    113.7000000  0.10480000 -0.02571600  0.00000000  0.00000000
     37.0300000  0.28306200 -0.07092400  0.00000000  0.00000000
     13.2700000  0.44871900 -0.16541100  0.00000000  0.00000000
      5.0250000  0.27095200 -0.11695500  0.00000000  0.00000000
      1.0130000  0.01545800  0.55736800  0.00000000  0.00000000
      0.3023000 -0.00258500  0.57275900  1.00000000  0.00000000
      0.0789600  0.00000000  0.00000000  0.00000000  1.00000000
# p functions
f   5    3    0
     17.7000000  0.04301800  0.00000000  0.00000000
      3.8540000  0.22891300  0.00000000  0.00000000
      1.0460000  0.50872800  0.00000000  0.00000000
      0.2753000  0.46053100  1.00000000  0.00000000
      0.0685600  0.00000000  0.00000000  1.00000000
# d functions
f   2    2    0
      1.1850000  1.00000000  0.00000000
      0.3320000  0.00000000  1.00000000

Specification of how to generate small component functions by kinetic balance. If no number or a 0 is given, the small component functions are generated both upwards and downwards, e.g p \rightarrow s, d. If the number is 1, the small component functions are generated upwards, e.g p \rightarrow d. If the number is 2, the small component functions are generated downwards, e.g p \rightarrow s.

For other values no small components functions generated.

The routine that generates the small component basis set by kinetic balance will delete duplicate functions.

MOLFDIR-type basis sets 

DIRAC
Water
MOLFDIR basis
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE MOLFBAS Oxygen-xyz.bas

The MOLFDIR basis set file(s) (here: Oxygen-xyz.bas) must be copied to the scratch area, for example using the pam script: 

pam ... -copy "Oxygen-xyz.bas ..."

Even-tempered basis sets (geometric progressions)

The exponents η are generated in an even-tempered series:

ηN − k + 1 = α βk − 1; k = 1, ..., N 

DIRAC
Water
even-tempered basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EVENTEMP 0.05 3.0 11 3 2 1 1
1..5
6..11
7..11
9..10

DIRAC
Water
even-tempered basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EVENTEMP 0.05 3.0 11 3 2 1 1
1..5
6..11
7..11
9..10

Highest angular quantum number l plus one. In this case it is 3, since we are using a spd basis set.

Number of blocks for each l-value. The memory requirements grow rapidly with the number of basis functions in a block (note for instance that four g functions actually are 60 basis functions, as there are 15 cartesian components of each g function). Memory requirements can therefore be reduced by splitting basis functions of the quantum number into different blocks. This will, however, decrease the performance of the integral calculation. 

DIRAC
Water
even-tempered basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE EVENTEMP 0.05 3.0 11 3 2 1 1
1..5
6..11
7..11
9..10

For each block the range of exponent used for that particular block is given. In this case the first block (s functions) will consist of exponents 1-5, the second block (s functions) of exponents 6-11, the third block (p functions) of exponents 7-11, and the last block (d functions) of exponents 9-10.

Well-tempered basis sets

The exponents η are generated in an well-tempered series:

ηN = α

ηN − k + 1 = ηN − k + 2 β \bigl[ 1 + γ \bigl( \frac{k}{N} \bigr) δ \bigr]; k = 1, ..., N 

DIRAC
Water
well-tempered basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE WELLTEMP 0.05 2.5 2.0 6.0 11 3 2 1 1
1..5
6..11
7..11
9..10

For the specification of blocks and exponent ranges see LARGE EVENTEMP.

Family basis sets

Input for basis sets where the same set of exponents are used for all functions. This is analogous to the well- and even-tempered basis sets except that the exponents are not calculated from a formula, but must be given in the file. These exponents may come from a basis set optimization with GRASP. 

DIRAC
Water
family basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE FAMILY   10    3    1    1    1
# exponents
 6665.0000000
 1000.0000000
  228.0000000
   64.7100000
   21.0600000
    7.4950000
    2.7970000
    0.5215000
    0.1596000
    0.0469000
# ranges
1..10
6..10
8..9

For the specification of blocks and exponent ranges see LARGE EVENTEMP.

Dual family basis sets

A basis set analogous to the family basis set, except one set of exponents are used for s, d, g, ... functions and another set is used for p, f, h, ... functions. 

DIRAC
Water
dual family basis set
C   2    2  X  Y
        8.    1
O      .0000000000        0.0000000000        -.2249058930
LARGE DUALFAMILY   10    5    3    1    1    1
# s, d, g, ...
 6665.0000000
 1000.0000000
  228.0000000
   64.7100000
   21.0600000
    7.4950000
    2.7970000
    0.5215000
    0.1596000
    0.0469000
# p, f, h, ...
    9.4390000
    2.0020000
    0.5456000
    0.1517000
    0.0404100
# ranges
1..10
1..5
8..9

For the specification of blocks and exponent ranges see LARGE EVENTEMP.

Retrieved from "http://wiki.chem.vu.nl/dirac/index.php/Manual:Basis:Large_component_basis_set"
Personal tools