0
$\begingroup$

We have the map $P(f,g)=f+g$, $C[0,1]^2 \rightarrow C[0,1]$, now I want to prove that it is continuous on $L^1$ metric.

The outline is find $\delta$, such that $d_1(d_1(f_1,g_1),d_1(f_2,g_2))<\delta$ implies $d_1(P(f_1,g_1),P(f_2,g_2))<\epsilon$.

I got $\left|\int_0^1|(f_1(t)-f_2(t))|\,\mathrm{d}t-\int_0^1|(g_1(t)-g_2(t))|\,\mathrm{d}t\right|$ for the left expression, and $\int_0^1|(f_1(t)-f_2(t))+(g_1(t)-g_2(t))|\,\mathrm{d}t$ for the right expression. However, using the triangle inequality, the inequality is opposite from the desired direction. Where is the mistake? It is because the definition for L1 is incorrect, or a different technique should be applied?

2 Answers 2

0

I reckon we equip $X =: C^0([0,1])$ with the norm $\Vert \,f\, \Vert = \int_{[0,1]} |\,f(s)|\,ds$ and $X^2$ with its natural product topology induced by $\Vert (f,g) \Vert_{\times} = \max\big\{\|f\|,\|g\|\big\}$. It suffices to prove sequential continuity. Let $(f_n,g_n) \rightarrow (f_0,g_0) \in X^2$, and we want to show that $P(f_n,g_n) \rightarrow P(f_0,g_0)$. Indeed:

\begin{align}\Vert P(f_0, g_0) - P(f_n,g_n) \Vert &= \int_{[0,1]}|f_n(s) - f_0(s) + g_n(s) - g_0(s)|ds \\&\leq \int_{[0,1]} |f_n(s) - f_0(s)| + |g_n(s) - g_0(s)|ds \\ &\leq \int_{[0,1]} |f_n(s) - f_0(s)|ds + \int_{[0,1]}|g_n(s) - g_0(s)|ds \end{align}

Since $\Vert (f_n - f_0,g_n-g_0) \Vert_{\times} \rightarrow 0$, the right-hand side goes to $0, finishing the proof.

  • 0
    Thanks. I found where I am wrong. I used L1 metric formula for points when combining the two distances.2017-02-21
0

First, you should be looking for $$ \lVert (f_1,g_1) - (f_2,g_2) \rVert_{L^1\times L^1} < \delta $$ not $$ d_1(d_1(f_1,g_1),d_1(f_2,g_2))<\delta. $$

Second, there is a sign error between the to integrals in your formula for the "left expression". If you fix that, the triangle equality will give you exactly what you need.

  • 0
    Thanks for correcting that distance formula, but can you tell me the explicit formula for the norm $L^1\times L^1$?2017-02-21
  • 0
    There is more than one way to define the norm on a product space, but in this case the easiest way to go is to take the sum of the norms, i.e. : $\lVert (f_1,g_1) - (f_2,g_2) \rVert_{L^1\times L^1}:= \lVert f_1 - f_2 \rVert_{L^1} + \lVert g_1 - g_2 \rVert_{L^1} $.2017-02-21
  • 0
    I see. I used a wrong subtraction formula...2017-02-21