A note on Blackwell's renewal theorem (Q5916232)

Let \(U(t)\) be the renewal function corresponding to a sequence of independent identically distributed random variables with common distribution \(F(t)\). Let \(m(t)\) be the integrated tail of \(F(t)\) and let \(Z(t)\) be the residual life (time) of the renewal process. The note contains two results: 1. Suppose \(F\) is nonlattice and has finite mean \(\mu\), then the following statements are equivalent: \(\lim_{t\to \infty}\{U(t+y)-U(t)\}=y/\mu\) and the distribution of \(Z(t)\) tends to some nondegenerate random value \(Z\). 2. In infinite mean case for \(a(t)\) such that \(\lim_{t\to \infty}a(t+y)/a(t)=1\) for any fixed \(y>0\), the following statements are equivalent: \(\lim_{t\to \infty}a(t)\{U(t+y)-U(t)\}=k(y)\) and \(\lim_{t\to \infty}a(t)\mathbb P\{Z(t)\leq y\}=v(y)\). Moreover, if one of the limits exists, then \(k(y)=\beta y\) and \(v(y)=\beta m(y)\) for some constant \(\beta\geq0\).