I agree. Unfortunately, the opposite is not necessarily true - LDA+U may converge well, and still be unphysical.
On 8/3/2010 11:36 AM, Laurence Marks wrote: > One important addendum. If LDA+U is unphysical, it may not converge > and nothing you do will make it! I ran into something like this some > time ago when I tried PBE0 for a Ni surface. > > Unless you run into a (rare) bug in Wien2k, failure to converge or > poor convergence is an indicator that the physics has been posed > wrong.