Ch · Chemistry

What Do Short-Range and Long-Range Corrections Mean in DFT?

Published By the 9to5Science editorial team

At a glance

  1. 1. Split the interaction

    Range-separated methods divide the Coulomb interaction into short-distance and long-distance parts using an error-function parameter.

  2. 2. Assign exchange

    A functional specifies how much Hartree–Fock and density-functional exchange it uses in each range.

  3. 3. Compare the limits

    Long-range-corrected methods often increase exact exchange at large distances; screened hybrids such as HSE reduce it there.

  4. 4. Choose for the problem

    The useful correction depends on the property and system; check the functional’s definition rather than relying on its label.

Advertisement

Title: What Do Short-Range and Long-Range Corrections Mean in DFT? Category: chemistry

Short answer: Short-range and long-range corrections in DFT refer to how a functional treats electron interactions at different interelectronic distances. Range-separated methods split the Coulomb interaction into short- and long-range pieces, then assign each piece a chosen amount of Hartree–Fock and density-functional exchange. “Long-range corrected” often means exact exchange grows at large distances.

---- ARTICLE BODY ----

In density-functional theory (DFT), short-range and long-range usually describe the distance between two electrons. A short-range correction changes how a functional treats nearby electron pairs; a long-range correction changes its treatment of pairs far apart. These terms do not name one universal correction formula: the precise meaning depends on the functional.

The distinction is useful because common approximations to exchange can behave poorly when electrons are far apart. Range-separated functionals address this by dividing the electron–electron Coulomb interaction into two parts and treating those parts differently.

How the Coulomb interaction is split

A common split uses the error function, erf⁡\operatorname{erf}, and its complement, erfc⁡\operatorname{erfc}:

1r=erf⁡(ωr)r+erfc⁡(ωr)r.\frac{1}{r}=\frac{\operatorname{erf}(\omega r)}{r}+\frac{\operatorname{erfc}(\omega r)}{r}.

Here, rr is the distance between electrons and ω\omega sets the length scale of the split. The erf⁡(ωr)/r\operatorname{erf}(\omega r)/r term is finite as rr approaches zero and becomes the long-range part at large rr. The erfc⁡(ωr)/r\operatorname{erfc}(\omega r)/r term contains the short-range singularity and decays at large rr. Increasing ω\omega shifts the crossover toward shorter distances.

The split itself is exact. The approximation enters when a functional chooses how to model exchange using this split; correlation is treated according to that functional’s separate prescription and is not necessarily split in the same way. In a hybrid functional, some exchange is calculated using Hartree–Fock (HF) exchange, while the rest comes from a density-functional approximation (DFA), such as a generalized-gradient approximation.

Schematically, a range-separated exchange energy can be written as

Ex=aSRExHF,SR+aLRExHF,LR+DFA exchange contributions.E_x = a_{\mathrm{SR}} E_x^{\mathrm{HF,SR}} + a_{\mathrm{LR}} E_x^{\mathrm{HF,LR}} + \text{DFA exchange contributions}.

The coefficients aSRa_{\mathrm{SR}} and aLRa_{\mathrm{LR}} specify the HF-exchange fractions at short and long range. This is a schematic description: actual functionals differ in how they define and combine the DFA contributions, and in how they treat correlation.

What “long-range corrected” usually means

Many long-range-corrected (LC) functionals use a relatively small or zero fraction of HF exchange at short range, then increase the fraction toward 100% at long range. LC-ω\omegaPBE is a familiar example. This behavior can improve the description of situations where separated parts of a system interact, including some charge-transfer excitations. It can also improve the long-distance behavior of the exchange potential.

Not every range-separated functional switches from one fraction to another in the same way. For example, CAM-B3LYP uses about 19% HF exchange at short range and 65% at long range. The “65%” is the long-range fraction, not a statement that all range-separated functionals use that value.

A different design is used by screened hybrids, notably HSE. HSE includes HF exchange at short range but screens it out at long range, leaving semilocal exchange there. This can be useful in calculations on solids, where unscreened long-range HF exchange may be computationally demanding or unsuitable for the intended approximation. Thus, “short-range corrected” may refer to a method that includes exact exchange only over short distances; it does not automatically mean the same thing as “long-range corrected.”

A numerical illustration of the split

Take ω=0.3 a0−1\omega=0.3\,a_0^{-1} and an electron separation of r=2 a0r=2\,a_0, where a0a_0 is the Bohr radius. Then ωr=0.6\omega r=0.6, and approximately erf⁡(0.6)=0.6039\operatorname{erf}(0.6)=0.6039 and erfc⁡(0.6)=0.3961\operatorname{erfc}(0.6)=0.3961. In atomic units, 1/r=0.51/r=0.5 Hartree, so the two pieces are

erf⁡(0.6)2≈0.302 Eh,erfc⁡(0.6)2≈0.198 Eh.\begin{aligned} \frac{\operatorname{erf}(0.6)}{2} &\approx 0.302\ E_h,\\ \frac{\operatorname{erfc}(0.6)}{2} &\approx 0.198\ E_h. \end{aligned}

They add to 0.500 Eh0.500\ E_h, the original Coulomb interaction. This example illustrates the mathematical split, not the full energy of a molecule: a DFT calculation integrates contributions over electron density and applies the functional’s exchange and correlation rules.

Why the distinction matters in practice

Semilocal functionals can suffer from self-interaction and delocalization errors. In some cases these errors make it too favorable to spread an electron over distant regions, or cause an incorrect energy variation as charge moves between separated fragments. Including more HF exchange at long range can reduce some of these errors and improve certain charge-transfer calculations.

That does not make long-range correction universally more accurate. Results depend on the molecule or material, the property being calculated, and the functional’s parameters. Long-range-corrected methods can change excitation energies and other results substantially, and their added exact exchange can increase computational cost. A functional that performs well for molecular charge transfer is not automatically the best choice for a solid or for every ground-state property.

Common mistakes

  • Treating the labels as a universal recipe. “Short-range corrected” and “long-range corrected” are not interchangeable names with one fixed formula. Check the HF-exchange fractions and the range-separation parameter for the specific functional.
  • Confusing distance range with energy range. Short and long refer to electron–electron separation, not to low- and high-energy electrons.
  • Assuming the Coulomb split changes the physical interaction. The two terms sum to the original 1/r1/r interaction; the approximation is in how the functional evaluates their exchange and correlation contributions.
  • Assuming more exact exchange always improves results. The effect is property- and system-dependent. Compare appropriate benchmarks or established methods for the system you are studying.

When reading a functional’s name or documentation, look for its short- and long-range HF-exchange fractions, its value of ω\omega, and whether the long-range interaction is corrected or screened. Those details explain what the method actually does.

References

  1. A new hybrid exchange-correlation functional using the Coulomb-attenuating method (CAM-B3LYP) — T. Yanai, D. P. Tew, and N. C. Handy, Chemical Physics Letters 393 (2004), 51–57
  2. Hybrid functionals based on a screened Coulomb potential — J. Heyd, G. E. Scuseria, and M. Ernzerhof, The Journal of Chemical Physics 118 (2003), 8207–8215

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.