Numerical solution of the electronic structure problem is playing an ever increasing rolein science and its applications, driven by relentless increases in computational power and algorithmic development. Many of these calculations currently involveKohn-Sham density functional theory, because of its present ability to handle hundreds if not thousands of atoms with useful (but not chemical) accuracy. These methods dominate in materials and nanoscience, and are being increasingly applied in chemical dynamics of complex systems, biology, other soft matter, and disordered systems. All such calculations are limited by the accuracy of present approximations or their inability to capture vital physics, such as asymptotic van der Waals forces. The approximations that dominate present use (generalized gradient approximation and hybrid functionals) were developed two decades ago. There are (too) many contenders for possible improvements, but none dominate or are universally applicable, nor will they in the forseeable future. Yet many applications require higher accuracy or new capabilities, such as for energy applications or biological molecules. The pressing need for improvement in specific applications has led to a spread of empiricism counter to the spirit of DFT.
The above comments apply even more to time-dependent DFT. What are the most important errors for which applications?Errors include adiabatic approximation, limitations of ground-state functionals,need for functionals beyond the exchange-correlation potential.Although remarkably successful for electronic spectra and response in a wide range of systems, present-day approximate functionals perform notoriously badly for many important applications: long-range charge-transfer excitations, molecular transport, electron dynamics in strong fields, to name a few.
Present conferences and workshops involving DFT are dominated by specific applications in specific fields. This workshop will be unique in recent times, by getting together DEVELOPERS of density functional methods from many different areas, and discussing the theoretical development, not the algorithms or specific applications.There will be tremendous input from experiment, in the form of successes and some spectacular failures of present methods reported. But more importantly, their significance will be analyzed in terms of the functionals.