Basis set extrapolation

Basis set extrapolation can be carried out for correlation consistent basis sets using

EXTRAPOLATE,BASIS=basislist,options

where basislist is a list of at least two basis sets separated by colons, e.g. AVTZ:AVQZ:AV5Z. Some extrapolation types need three or more basis sets, others only two. The default is to use $n^{-3}$ extrapolation of the correlation energies, and in this case two subsequent basis sets and the corresponding energies are needed. The default is not to extrapolate the reference (HF) energies; the value obtained with the largest basis set is taken as reference energy for the CBS estimate. However, extrapolation of the reference is also possible by specifying the METHOD_R option.

The simplest way to perform extraplations for standard methods like MP2 or CCSD(T) is to use, e.g.

***,H2O
memory,32,m
gthresh,energy=1.d-8

r =   0.9572 ang, theta =  104.52
geometry={O;
          H1,O,r;
          H2,O,r,H1,theta}

basis=avtz
hf
ccsd(t)
extrapolate,basis=avqz:av5z

table,basissets,energr,energy-energr,energy
head,basis,ehf,ecorr,etot

This will perform the first calculation with AVTZ basis, and then compute the estimated basis set limit using the AVQZ and AV5Z basis sets. The correlation energy obtained in the calculation that is performed immediately before the extrapolate command will be extrapolated (in this case the CCSD(T) energy), and the necessary sequence of calculations [here HF;CCSD(T)] will be automatically carried out.

The resulting energies are returned in variables ENERGR (reference energies), ENERGY (total energies), and ENERGD (Davidson corrected energy if available); the corresponding basis sets are returned in variable BASISSETS. The results can be printed, e.g., in a table as shown above, or used otherwise. The above input produces the table

 BASIS       EHF           ECORR          ETOT
 AVQZ    -76.06600082   -0.29758099   -76.36358181
 AV5Z    -76.06732050   -0.30297495   -76.37029545
 CBS     -76.06732050   -0.30863418   -76.37595468

The extrapolated total energy is also returned in variable ECBS (ECBSD for Davidson corrected energy if available).

In order to extrapolate the HF energy as well (using exponential extrapolation), three energies are needed. One can modify the input as follows:

extrapolate,basis=avtz:avqz:av5z,method_r=ex1,npc=2

method_r determines the method for extrapolating the reference energy (in this case a single exponential); npc=2 means that only the last two energies should be used to extrapolate the correlation energy (by default, a least square fit to all given energies is used). This yields

 BASIS           EREF            ECORR           ETOT
 AVTZ        -76.06061330     -0.28167606    -76.34228936
 AVQZ        -76.06600082     -0.29758099    -76.36358180
 AV5Z        -76.06732050     -0.30297495    -76.37029545
 CBS         -76.06774863     -0.30863419    -76.37638283

Rather than using the default procedure as above, one can also specify a procedure used to carry out the energy calculation, e.g.

extrapolate,basis=avtz:avqz:av5z,proc=runccsd, method_r=ex1,npc=2}

procedure runccsd
hf
ccsd(t)
endproc

Alternatively, the energies can be provided via variables EREF, ECORR, ETOT etc. These must be vectors, holding as many values as basis sets are given.

The possible options and extrapolation methods are:

  • BASIS=basissets Specify as set of correlation consistent basis sets, separated by colons.
  • PROC=procname Specify a procedure to run the energy calculations
  • STARTCMD=command Start command for the energy calculations: the sequence of commands from STARTCMD and the current EXTRAPOLATE is run. STARTCMD must come before the extrapolate command in the input.
  • METHOD=key Specifies a keyword to define the extrapolation function, see section extrapolation functionals.
  • METHOD_C=key Specifies a keyword to define the extrapolation function for the correlation energy, see section extrapolation functionals.
  • METHOD_R=key Specifies a keyword to define the extrapolation function for the reference energy, see section extrapolation functionals.
  • VARIABLE=name Specifies a variable name; this variable should contain the energies to be extrapolated.
  • ETOT=variable Provide the total energies in variable (a vector with the same number of energies as basis sets are given) If only ETOT but not EREF is given, the total energy is extrapolated.
  • EREF=variable Provide the reference energies to be extrapolated in variable (a vector with the same number of energies as basis sets are given)
  • ECORR=variable Provide the correlation energies to be extrapolated in variable (a vector with the same number of energies as basis sets are given)
  • ECORRD=variable Provide the Davidson corrected correlation energies to be extrapolated in variable (a vector with the same number of energies as basis sets are given). If both ECORR and ECORRD are given, both will be extrapolated.
  • MINB=number First basis set to be used for extrapolation (default 1)
  • MAXB=number Last basis set to be used for extrapolation (default number of basis sets)
  • NPR=number If given, the last NPR values are used for extrapolating the reference energy. NPR must be smaller or equal to the number of basis sets.
  • NPC=number If given, the last NPC values are used for extrapolating the reference energy. NPC must be smaller or equal to the number of basis sets.
  • XR=array Provide a vector of exponents to be used for defining the extrapolation functional for the reference energy when using the LX functional.
  • XC=array Provide a vector of exponents to be used for defining the extrapolation functional for the correlation energy when using the LX functional.
  • PR=array Provide the constant $p$ to be used for defining the extrapolation functional for the reference energy.
  • PC=array Provide the constant $p$ to be used for defining the extrapolation functional for the correlation energy.

The extrapolation functional is chosen by a keyword with the METHOD, METHOD_R, and/or METHOD_C options. The default functional is L3. In the following, $n$ is the cardinal number of the basis set (e.g., 2 for VDZ, 3 for VTZ etc), and $x$ is an arbitrary number. $p$ is a constant given either by the PR or PC options (default $p=0$). X is a number or a vector given either by the XR or XC options (only for LX; $nx$ is the number of elements provided in X). $A$, $B$, $A_i$ are the fitting coefficients that are optioized by least-squares fits.

  • L$x$ $E_{n} = E_{\tt CBS} + A \cdot (n+p)^{-x}$
  • LH$x$ $E_{n} = E_{\tt CBS} + A \cdot (n+\frac{1}{2})^{-x}$
  • LX $E_{n} = E_{\tt CBS} + \sum_{i=1}^{nx} A_i \cdot (n+p)^{-x(i)}$
  • EX1 $E_{n} =E_{\tt CBS}+A\cdot \exp(-C\cdot n)$
  • EX2 $E_{n} =E_{\tt CBS}+A\cdot \exp(-(n-1))+B\cdot\exp(-(n-1)^2)$
  • KM Two-point formula for extrapolating the HF reference energy, as proposed by A. Karton and J. M. L. Martin, Theor. Chem. Acc. 115, 330 (2006): $E_{\rm HF,n}=E_{\rm HF,CBS} +A (n+1)\cdot \exp(-9 \sqrt{n})$. Use METHOD_R=KM for this.

The following example shows various possibilities for extrapolation:

examples/h2o_extrapolate_ccsd.inp
***,h2o

gthresh,energy=1.d-9
basis=avtz

r =   0.9572 ang, theta =  104.52
geometry={!nosym
          O;
          H1,O,r;
          H2,O,r,H1,theta}

hf
{ccsd(t)}
text,compute energies, extrapolate reference energy using EX1 and correlation energy using L3
extrapolate,basis=avtz:avqz:av5z,method_c=l3,method_r=ex1,npc=2

ehf=energr(1:3)
etot=energy(1:3)

text,extrapolate total energy using EX2
extrapolate,basis=avtz:avqz:av5z,etot=etot,method=ex2

text,extrapolate reference energy by EX1 and correlation energy by EX2
extrapolate,basis=avtz:avqz:av5z,etot=etot,method_c=ex2,eref=ehf,method_r=ex1

text,extrapolate reference energy by EX1 and correlation energy by LH3
extrapolate,basis=avtz:avqz:av5z,etot=etot,method_c=LH3,eref=ehf,method_r=ex1,npc=2

text,extrapolate reference energy by EX1 and correlation energy by LX
extrapolate,basis=avtz:avqz:av5z,etot=etot,method_c=LX,eref=ehf,method_r=ex1,xc=[3,4],pc=0.5

The second example shows extrapolations of MRCI energies. In this case both the MRCI and the MRCI+Q energies are extrapolated.

examples/h2o_extrapolate_mrci.inp
***,h2o

gthresh,energy=1.d-9
basis=avtz

r =   0.9572 ang, theta =  104.52
geometry={
          O;
          H1,O,r;
          H2,O,r,H1,theta}

hf
multi
mrci
text,Compute energies, extrapolate reference energy using EX1 and correlation energy using L3;
text,The Davidson corrected energy is also extraplated
extrapolate,basis=avtz:avqz:av5z,method_c=l3,method_r=ex1,npc=2

emc=energr
ecorr_mrci=energy-emc
ecorr_mrciq=energd-emc

text,Extrapolate reference energy by EX1 and correlation energy by LH3
text,The Davidson corrected energy is also extraplated
extrapolate,basis=avtz:avqz:av5z,ecorr=ecorr_mrci,ecorrd=ecorr_mrciq,method_c=LH3,eref=emc,method_r=ex1,npc=2

Geometry optimizations are possible by using numerical gradients obtained from extrapolated energies. Analytical energy gradients cannot be used.

The following possibilities exist:

1.) If OPTG directly follows the EXTRAPOLATE command, the extrapolated energy is optimized automatically (only variable settings may occur between EXTRAPOLATE and OPTG).

Examples:

Extrapolating the energy for the last command:

examples/h2o_extrapol_opt1.inp
geometry={o;h1,o,r;h2,o,r,h1,theta}
theta=102
r=0.96 ang
basis=vtz

hf
ccsd(t)
extrapolate,basis=vtz:vqz

optg

Extrapolating the energy computed in a procedure:

examples/h2o_extrapol_opt2.inp
geometry={o;h1,o,r;h2,o,r,h1,theta}
theta=102
r=0.96 ang

proc ccsdt
hf
ccsd(t)
endproc

extrapolate,basis=vtz:vqz,proc=ccsdt

optg

Note that this is not possible if EXTRAPOLATE gets the input energies from variables.

2.) Using a procedure for the extrapolation:

By default, variable ECBS is optimized, but other variables (e.g. ECBSD) can be specified using the VARIABLE option on the OPTG command.

examples/h2o_extrapol_opt3.inp
geometry={o;h1,o,r;h2,o,r,h1,theta}
theta=102
r=0.96 ang
basis=vtz

proc cbs34
hf
ccsd(t)
extrapolate,basis=vtz:vqz
endproc

optg,variable=ecbs,proc=cbs34
examples/h2o_extrapol_opt4.inp
geometry={o;h1,o,r;h2,o,r,h1,theta}
theta=102
r=0.96 ang

proc cbs34
basis=vtz
hf
ccsd(t)
eref(1)=energr
ecc(1)=energy

basis=vqz
hf
ccsd(t)
eref(2)=energr
ecc(2)=energy
extrapolate,basis=vtz:vqz,eref=eref,etot=ecc
endproc

optg,variable=ecbs,proc=cbs34

This is possible by defining the extrapolation in a procedure:

examples/h2o_extrapol_freq.inp
geometry={o;h1,o,r;h2,o,r,h1,theta}
theta=102
r=0.96 ang
basis=vtz

proc cbs34
hf
ccsd(t)
extrapolate,basis=vtz:vqz
endproc

optg,variable=ecbs,proc=cbs34
freq,variable=ecbs,proc=cbs34