Semilocal density functionals and constraint satisfaction

We discuss six questions related to the recent strongly constrained and appropriately normed (SCAN) meta-generalized gradient approximation (meta-GGA): (1) When and why should a semilocal approximation to the density functional for the exchange-correlation energy be accurate? (2) What is the right dimensionless ingredient for a meta-GGA, and why? (3) In the construction of density functional approximations, should we satisfy more or fewer exact constraints? (4) Is there a tight lower bound on the exchange energy for all spin-unpolarized densities? (5) Should a semilocal approx- imation yield any intermediate-range van der Waals interaction? (6) Do semilocal functionals make consistent predictions for the energy differences between different molecules (and thus presumably for reaction and formation energies)? (c) 2016 Wiley Periodicals, Inc.