Abstract
The determination of an approximate wavefunction in the form ∑i=1
n Aiexp(-bir2), 1≤n≤10, by minimization of the variance, ∫|(H-E)ψ| 2dτ, is tested on the H atom. Obstacles to using other definitions of variance are discussed. Results are less favorable than those obtained by minimization of 〈H〉, but improve rapidly as n increases.