Interactive · 1 question
The Bolzano-Weierstrass approach involves a theorem in analysis called the Bolzano-Weierstrass theorem. Using this theorem, we'll prove the Boundedness Theorem, and then use the Boundedness Theorem to prove the Extreme Value Theorem.
Theorem (Bolzano-Weierstrass): Every bounded sequence in has a convergent subsequence.
We will not prove this theorem (see any real analysis textbook for the proof), but let's look at what this theorem means. Let's say we have a bounded sequence of real numbers . Bounded means that there is some such that , for all . The Bolzano-Weierstrass theorem then says that there is some subsequence of that converges to some . A subsequence of is what is sounds like - you take and just pick some of those (in order!) to be your subsequence. A simple example is picking all the even terms, or all the odd terms.
Example 1: Consider the sequence where . This is the sequence , which is bounded, with say, . This sequence does not converge (try showing this by contradiction, using the definition of a limit of a sequence and the triangle inequality. The key fact is that for every , there exist with and .) However, it does have a subsequence that converges. One such example is all the odd terms, to make the sequence , which clearly converges to .
Example 2: To see why bounded is necessary in the theorem, consider the sequence . Here , and diverges to . But every subsequence of also diverges to , so there is no subsequence that converges to .
Now it's time for the Boundedness Theorem.
What were the conditions in the theorem to guarantee that is bounded?
Theorem (Boundedness Theorem): A continuous function is bounded.
Proof: Suppose is not bounded. Then for every , there exists some with . The sequence is bounded since it is inside , so has a convergent subsequence, call it that converges to some . Now consider . Since is continuous, we have . But since , we have , a contradiction. Thus must be bounded.
Theorem (Extreme Value Theorem): A continuous function attains its maximum and minimum.
Proof: By the Boundedness Theorem, is bounded. Consider the supremum (least upper bound) of . We need to show that there is some with , i.e. attains this value. Since is the least upper bound, we can form a sequence such that (since this is not an upper bound for . Moreover, implies that . Now consider . By the Bolzano-Weierstrass theorem, this implies that has a convergent subsequence . By continuity of , then , so attains its maximum .
For the minimum, consider .
Remark: Why did we need to use the Bolzano-Weierstrass theorem? That is, why doesn't the conclusion finish the proof?
The conclusion doesn't tell us what the actually do. For example, consider on . Look at the sequence . The idea is that oscillates between approaching and . Observe that is simply , which goes to as . However, the sequence does not converge to anything. We need the Bolzano-Weierstrass theorem to ensure that we can get a convergent subsequence, and then we use the limit of that convergent subsequence to get the such that . Without having a limit of some sequence, we can't get that .