On page 348 of Infinite Dimensional Dynamical Systems in Mechanics and Physics by Roger Temam, there is something I don't understand. I abstracted the question as follow: Let $T:H \rightarrow H$ be a bounded self-adjoint operator on Hilbert space $H$. If $T$ is nonnegative and have bounded inverse $T^{-1}$ then why $$ \inf_{\| \phi \|=1} \langle T\phi , \phi \rangle >0. $$ is valid?
I found a similar question here Bounded Self-adjoint Operator on Hilbert Space, however it only provide answer when $T$ is positive, is it true for nonnegative operator as well?
Let $m:=\inf_{\| \phi \|=1} \langle T\phi , \phi \rangle $ and $M:=\sup_{\| \phi \|=1} \langle T\phi , \phi \rangle $.
It is well known that for the spectrum of $T$ we have
$$ \sigma(T) \subseteq [m,M].$$
Suppose that that $m=0$, then $0 \in \sigma(T),$ a contradiction, since $T$ has a bounded inverse.