Superadditivity and monotonicity of some functionals associated with the Hermite-Hadamard inequality for convex functions in linear spaces (Q691503)

If \(f\) is convex then the functional \[ \Psi_f(x,y;t)=(1-t)f(x) +tf(y) -f\bigl((1-t)x +ty\bigr),\;0\leq t\leq 1, \] is non-negative. In this paper it is shown that in addition \(\Psi_f(\cdot,\cdot;t)\) is super-additive and increasing as a function of the interval. Further it is shown that the same is true of the functionals \[ \Gamma_f(x,y;t)=f(x) + f(y) -f\bigl((1-t)x+ty\bigr)-f\bigl(tx+(1-t)y\bigr), \] \[ \Theta(x,y) ={{f(x) + f(y)}\over {2}}-\int_0^1f\bigl((1-t)x +ty\bigr)dt. \] Several interesting applications of these facts to norms and means are then made.