Density functional theory
A wavefunction of electrons is a function of coordinates. For benzene, that is 126 dimensions. The electron density, how many electrons are where, depends on just three. In 1964 Pierre Hohenberg and Walter Kohn proved that the density alone determines everything about a molecule’s ground state. Density functional theory (DFT), built on that theorem, is now the workhorse of chemistry, materials science and drug design, used in tens of thousands of papers a year. Kohn shared the 1998 Nobel Prize in Chemistry. DFT’s failures point to exactly where quantum computers might help.
On the plate: what happens to the energy as you add a fraction of an electron.
Hohenberg–Kohn and Kohn–Sham
The first Hohenberg–Kohn theorem: the external potential, and hence the whole Hamiltonian, is fixed (up to a constant) by the ground-state density . The proof is short: if two potentials gave the same density, applying the variational principle (Chapter 46) to each ground state with the other’s Hamiltonian gives a contradiction. The second theorem: the true density minimizes a universal energy functional . Nobody knows that functional exactly; it is QMA-hard to compute in general (Chapter 129).
In 1965 Kohn and Lu Jeu Sham made it practical: replace the interacting electrons with fictitious non-interacting ones that have the same density, moving in an effective potential:
Everything hard is swept into the exchange–correlation potential , which must be approximated. Established
Delocalization error
The exact energy, as a function of a fractional number of electrons , is a series of straight lines between integers. Approximate functionals bend those lines: common ones are convex, which over-stabilizes electrons spread out across several fragments, the delocalization error; Hartree–Fock bends the other way. Set the curvature to 0.20 (positive like typical density functionals, negative like Hartree–Fock). Aron Cohen, Paula Mori-Sánchez and Weitao Yang showed in 2008 that this one flaw explains many DFT failures: underestimated reaction barriers and band gaps, overly delocalized charge in transition states, and the wrong dissociation of H₂⁺, where a single electron interacts with its own density, the self-interaction error. Established
Jacob’s ladder, and where DFT fails
John Perdew organized functionals in 2001 as rungs of a ladder toward “chemical heaven”: the local density approximation (LDA) uses only the density at each point; generalized-gradient approximations (GGA, such as PBE) add its gradient; meta-GGAs add the kinetic-energy density; hybrids (such as B3LYP) mix in a fraction of exact Hartree–Fock exchange; double hybrids add perturbative correlation. Each rung usually costs more and helps more, though not always. DFT handles most organic biochemistry well. It struggles where one configuration does not dominate: transition-metal clusters, such as the iron–sulfur clusters of electron-transport proteins and the FeMo cofactor of nitrogenase, where different functionals disagree by tens of kcal/mol on which spin state is lowest. The spread between functionals is itself an honest error bar. Those strongly correlated active sites are the leading targets for quantum computers in Part 5. Established
What you now have
The Hohenberg–Kohn theorems and their proof idea, the Kohn–Sham equations, the exchange–correlation functional, the piecewise-linear exact energy and delocalization error, self-interaction, Jacob’s ladder, and the strongly correlated systems where DFT fails.
· · ·
Three questions to carry with you.
And one to work out. Benzene has 42 electrons. How many spatial coordinates does its wavefunction depend on?
Sources: P. Hohenberg and W. Kohn, Phys. Rev. 136, B864 (1964). W. Kohn and L. J. Sham, Phys. Rev. 140, A1133 (1965). J. P. Perdew and K. Schmidt, AIP Conf. Proc. 577, 1 (2001). A. J. Cohen, P. Mori-Sánchez and W. Yang, Science 321, 792 (2008). J. P. Perdew et al., Phys. Rev. Lett. 49, 1691 (1982) (piecewise linearity).