Kontinuitet og kompakthet
Posted: 04/10-2011 21:56
Hei.
Setter pris på om noen kan vurdere om jeg har gjort et bevis korrekt. Oppgaven lyder:
If [tex]f[/tex] is defined on [tex]E[/tex], the graph of [tex]f[/tex] is the set of points [tex](x, f(x))[/tex], for [tex]x \in E[/tex]. In particular, if [tex]E[/tex] is a set of real numbers, and [tex]f[/tex] is real-valued, the graph of [tex]f[/tex] is a subset of the plane.
Suppose [tex]E[/tex] is compact, and prove that [tex]f[/tex] is continuous on [tex]E[/tex] if and only if its graph is compact.
OK. I forbindelse med oppgaven bruker jeg følgende teoremer fra Rudin:
Theorem 4.8: A mapping [tex]f[/tex] of a metric space [tex]X[/tex] into a metric space [tex]Y[/tex] is continuous on [tex]X[/tex] if and only if [tex]f^{-1}(V)[/tex] is open in [tex]X[/tex] for every open set [tex]V[/tex] in [tex]Y[/tex].
Theorem 4.14: Suppose [tex]f[/tex] is a continuous mapping of a compact metric space [tex]X[/tex] into a metric space [tex]Y[/tex]. Then [tex]f(X)[/tex] is compact.
Theorem 4.19: Let [tex]f[/tex] be a continuous mapping of a compact metric space [tex]X[/tex] into a metric space [tex]Y[/tex]. Then [tex]f[/tex] is uniformly continuous on [tex]X[/tex].
OK. Løsning: Begynner med å bevise i forward direction:
Given the compact set [tex]E[/tex]. Assume that the mapping [tex]x \rightarrow (x, f(x))[/tex] is continuous. It then follows from theorem 4.14 that the graph also must be compact.
Backwards direction:
Proof by contradiction -
Suppose [tex]E[/tex] is compact and assume we have a graph [tex](x, f(x))[/tex] for [tex]x \in E[/tex] that is not compact. Accoring to the Heine-Borel theorem this means that the graph is either open or unbounded.
Suppose first that the graph [tex](x, f(x))[/tex] is open. Then, according to theorem 4.8, the graph's inverse image must also be open for the function to be continuous. However, since we assume that [tex]E[/tex] is compact this must mean that the function can not be continuous. Hence, the graph must be closed for [tex]f[/tex] to be continuous.
Now suppose that the graph [tex](x, f(x))[/tex] is unbounded. Then the graph is not uniformly continuous since we can find a [tex]\delta > 0[/tex] and [tex]\epsilon > 0[/tex] such that for two values [tex]x[/tex] and [tex]y[/tex] in [tex]E[/tex] we have that [tex]|x - y| < \delta[/tex] whereas [tex]|f(x) - f(y)| \geq \epsilon[/tex]. According to theorem 4.19 this graph is therefore not continuous either. Hence, the graph must be bounded for [tex]f[/tex] to be continuous.
Since we have shown that the graph must be be both closed and bounded for [tex]f[/tex] to be continuous, the graph must be compact.
QED
Setter stor pris på innspill/kommentarer!
Setter pris på om noen kan vurdere om jeg har gjort et bevis korrekt. Oppgaven lyder:
If [tex]f[/tex] is defined on [tex]E[/tex], the graph of [tex]f[/tex] is the set of points [tex](x, f(x))[/tex], for [tex]x \in E[/tex]. In particular, if [tex]E[/tex] is a set of real numbers, and [tex]f[/tex] is real-valued, the graph of [tex]f[/tex] is a subset of the plane.
Suppose [tex]E[/tex] is compact, and prove that [tex]f[/tex] is continuous on [tex]E[/tex] if and only if its graph is compact.
OK. I forbindelse med oppgaven bruker jeg følgende teoremer fra Rudin:
Theorem 4.8: A mapping [tex]f[/tex] of a metric space [tex]X[/tex] into a metric space [tex]Y[/tex] is continuous on [tex]X[/tex] if and only if [tex]f^{-1}(V)[/tex] is open in [tex]X[/tex] for every open set [tex]V[/tex] in [tex]Y[/tex].
Theorem 4.14: Suppose [tex]f[/tex] is a continuous mapping of a compact metric space [tex]X[/tex] into a metric space [tex]Y[/tex]. Then [tex]f(X)[/tex] is compact.
Theorem 4.19: Let [tex]f[/tex] be a continuous mapping of a compact metric space [tex]X[/tex] into a metric space [tex]Y[/tex]. Then [tex]f[/tex] is uniformly continuous on [tex]X[/tex].
OK. Løsning: Begynner med å bevise i forward direction:
Given the compact set [tex]E[/tex]. Assume that the mapping [tex]x \rightarrow (x, f(x))[/tex] is continuous. It then follows from theorem 4.14 that the graph also must be compact.
Backwards direction:
Proof by contradiction -
Suppose [tex]E[/tex] is compact and assume we have a graph [tex](x, f(x))[/tex] for [tex]x \in E[/tex] that is not compact. Accoring to the Heine-Borel theorem this means that the graph is either open or unbounded.
Suppose first that the graph [tex](x, f(x))[/tex] is open. Then, according to theorem 4.8, the graph's inverse image must also be open for the function to be continuous. However, since we assume that [tex]E[/tex] is compact this must mean that the function can not be continuous. Hence, the graph must be closed for [tex]f[/tex] to be continuous.
Now suppose that the graph [tex](x, f(x))[/tex] is unbounded. Then the graph is not uniformly continuous since we can find a [tex]\delta > 0[/tex] and [tex]\epsilon > 0[/tex] such that for two values [tex]x[/tex] and [tex]y[/tex] in [tex]E[/tex] we have that [tex]|x - y| < \delta[/tex] whereas [tex]|f(x) - f(y)| \geq \epsilon[/tex]. According to theorem 4.19 this graph is therefore not continuous either. Hence, the graph must be bounded for [tex]f[/tex] to be continuous.
Since we have shown that the graph must be be both closed and bounded for [tex]f[/tex] to be continuous, the graph must be compact.
QED
Setter stor pris på innspill/kommentarer!