Given \(g_n:[0,1]\to\mathbb{R}\) and \(g:[0,1]\to\mathbb{R}\), assume that there is a number \(M\) such that \(|g_n(x)\leqslant M|\) for all \(n\) and \(g(x)\leqslant M\) for all \(x\in[0,1]\). Moreover, assume that \(g_n\) and \(g\) are all integrable on \([0,1]\) and \(g_n(x)\to g(x)\) pointwise on \([0,1]\). With the above assumptions, show that if \(g_n\to g\) uniformly on \([\delta,1]\) for any \(0<\delta<1\), then\[\lim_{n\to\infty}\int_{0}^{1}g_n=\int_{0}^{1}g.\]
You do not really need the last assumption to show the last limit. The limit follows from the bounded convergence theorem. See http://mathworld.wolfram.com/LebesguesDominatedConvergenceTheorem.html
May be they are giving you this assumption to make the proof easier.
Here is a sketch of the proof with that assumption. \[ \int_0^{\delta} \left |g_n(t) - g(t) \right| dt\le 2M \int_0^{\delta}dt= 2 M \delta\\ \] Let \[ \epsilon >0 \] and choose \[ \delta=\min(\frac { \epsilon} { 4 M}, \frac 12) \] and choose N> so for n> M we have \[\left| g_n(t)- g(t) \right| \le \frac \epsilon {4 (1-\delta)} \] for any t in \[ [ \delta, 1]\le\\ \] \[ \int_0^{1} \left |g_n(t) - g(t) \right| dt=\int_0^{\delta} \left |g_n(t) - g(t) \right|dt+ \int_{\delta}^1 \left |g_n(t) - g(t) \right|dt\le2 M \delta + \frac \epsilon {4 (1-\delta)}(1-\delta)\\ \frac \epsilon 2 + \frac \epsilon 4 < \epsilon \]
abs Int <int abs so it is going to zero.
Join our real-time social learning platform and learn together with your friends!