Since the inception of Density Functional Theory (DFT) the remarkable success of the Local Density Approximation (LDA) has been difficult to improve in a systematic way. Originally Hohenberg, Kohn and Sham introduced LDA as the first term in a gradient expansion of the exchange-correlation energy functional. It success in a wide variety of systems, such as atoms molecules and solids[2, 3], was somewhat surprising, since the density gradients are not small. The accuracy of LDA was attributed to the sum rules which it satisfies and to the range of validity of the small gradient approximation being larger than expected[5, 6]. Well defined gradient expansions were carried out, however the most accurate numerical results for real systems require either a semi-empirical approach or a detailed model for the exchange-correlation hole. A large number of exact constraints have also been placed upon the possible functionals which are beginning to lead to more systematic improvements.
|Translated title of the contribution||Valence density functionals|
|Title of host publication||Density Functional Theory|
|Editors||Eberhard K U Gross, Reiner M Dreizler|
|Number of pages||11|
|Publication status||Published - 1995|
|Name||NATO Science Series B: Physics|