David C. Langreth died from complications following pneumonia on May 27, 2011. He was an outstanding condensed matter theorist as well as a friend, mentor, and role model for many in density functional theory, surface physics, many-body physics, scattering theory, and transport theory. Although modest and collaborative by nature, he was an original and very deep thinker. His work underpins many of the methodological developments of modern electronic structure theory, especially density functional theory, and deserves to be better known to the many users of those methods.
David was born in 1937. He received his B.S. from Yale in 1959 and his Ph.D. from the University of Illinois in 1964, where he was a student of Leo Kadanoff. At Illinois he received a rigorous education in diagrammatic many-body theory, which he used brilliantly throughout his research career. He was a research associate at the University of Chicago, and later at Cornell mainly with Neil Ashcroft, before beginning a faculty position in Physics at Rutgers in 1967. He remained at Rutgers for the rest of his career but made many scientific visits to institutions such as the Niels Bohr Institute, UC Santa Barbara, and Chalmers University, where he developed strong collaborative relationships. He was elected a Fellow of the APS in 1981, was the recipient of the Rutgers Board of Trustees Prize for Excellence in Research 1999, and received an honorary doctoral degree from Chalmers University of Technology in 2004.
David made many pioneering contributions to density functional theory (DFT), first proposed by Hohenberg, Kohn, and Sham in the sixties and now widely adopted as part of the 'standard model' of electronic-structure theory. Initially the unexpected success of DFT in the local-density approximation (LDA) was quite mysterious. In the mid-70's Langreth and Perdew derived an adiabatic-connection fluctuation-dissipation theorem at fixed inhomogeneous electron density, providing a formally exact expression for the exchange-correlation (XC) energy. These authors, along with Gunnarsson and Lundqvist, identified the density of the XC hole around an electron and derived its sum rules.These rules were also found to be satisfied by the LDA, thus largely resolving the mystery. Shortly thereafter, Langreth and Perdew constructed the random-phase approximation in the context of DFT, which has recently been implemented for molecules and solids, and carried out pioneering studies of surface energies in the DFT context.
Perhaps most importantly, Langreth and collaborators were at the forefront of the development of the generalized gradient approximation (GGA) to the XC energy. It was clear to many that the LDA might be improved upon by including gradient terms in the density functional, but initial attempts via a perturbation expansion in the density gradient proved unsuccessful. In 1980 Langreth and Perdew showed that the second-order gradient expansion violates the XC sum rule and that applying a reciprocal-space cutoff could restore the sum rule. Shortly thereafter, Langreth and Mehl used this idea to propose the first practical GGA, and further refinements by others over the next decade led to the modern GGA functionals in use today. The improved accuracy of the GGA was the critical factor leading to the growing acceptance of DFT methods in the chemistry community, as recognized via the award of the 1998 Nobel Prize in Chemistry to Walter Kohn and John Pople.
David also applied his deep knowledge of many-body theory to many other problems. In 1966 he proved a sum rule for the Anderson impurity model that became known as the Friedel-Langreth relation in that context. With Ashcroft at Cornell, he showed that many liquid metals and alloys can be described by pseudopotential perturbation theory using ion-ion structure factors as they emerged from advances in the microscopic theory of classical fluids. He solved a plasmaron problem exactly in 1970, explaining threshold behavior in the X-ray spectra of metals. In 1976 he showed how the Keldysh and Kadanoff-Baym treatments of non-equilibrium Green’s functions were simply related, and proved the Langreth theorem for extracting physical response functions. In later years he delighted in applying many-body techniques to new areas, especially surfaces. For example, he used slave boson methods to predict the charge-state distributions of ions scattering from surfaces.
David spent almost a quarter century mulling over the absence of dispersion forces in density functional approximations with his long-time sailing companion Bengt Lundqvist. This work came to fruition in the last decade, as they led the development of DFT methods capable of treating van der Waals (vdW) interactions. With Lundqvist and other collaborators, David developed fully nonlocal yet practical XC functionals suitable for treating vdW interactions, making use of the same adiabatic-connection approach developed much earlier in his career. This extended the domain of DFT to many biological and soft-matter systems for the first time, and is now an extremely active area of research.
Anyone who was fortunate enough to interact with David will miss his quiet manner and his thoughtful insights. Three of us were junior colleagues, and benefited enormously from his mentorship and guidance, both in our science and our careers. But perhaps what we will miss the most is his booming laugh, which could be heard for miles, and always brightened any scientific discussion. Neil Ashcroft (Cornell University) Kieron Burke (U.C. Irvine) John P. Perdew (Tulane University) David Vanderbilt (Rutgers University)