Lecture 27:

The Cauchy Criterion

Example. Define the sequence \(S = (s_{n})\) by

\[s_{n} = \sum_{k=1}^{n}\frac{1}{k}\quad\text{for all }n\in\N.\]

Notice that

\[s_{n+1}-s_{n} = \left(1+\frac{1}{2}+\frac{1}{3}+\cdots+\frac{1}{n}+\frac{1}{n+1}\right) - \left(1+\frac{1}{2}+\frac{1}{3}+\cdots+\frac{1}{n}\right)\]

\[ = \frac{1}{n+1}\]

So, for very large \(n\), the terms \(s_{n}\) and \(s_{n+1}\) are very close to each other. 

But, is \((s_{n})\) Cauchy?

Indeed, given any \(\varepsilon>0,\) no matter how small, we can find \(n\in\N\) large enough that \(\frac{1}{n+1}<\varepsilon,\) and hence for any \(m>n\) we have

\[s_{m+1}-s_{m}<\varepsilon.\]

Theorem. If \((x_{n})\) is a Cauchy sequence, then \((x_{n})\) is bounded. 

 

Proof. Since \((x_{n})\) is Cauchy, there is some \(H\in\N\) such that

\[|x_{n}-x_{m}|<1\quad\text{for all }n,m\geq H\]

Note that if \(n\geq H,\) then

\[|x_{H} - x_{n}|<1,\]

and hence

\[-1+x_{H}<x_{n}<1+x_{H}.\]

Set

\[M = \max\{|x_{1}|,|x_{2}|,\ldots,|x_{H-1}|,|1+x_{H}|,|-1+x_{H}|\}.\]

Then we have

\[|x_{n}|\leq M\quad\text{for all }n\in\N,\]

that is, \((x_{n})\) is bounded. \(\Box\)

Example. Define the sequence \(S = (s_{n})\) by

\[s_{n} = \sum_{k=1}^{n}\frac{1}{k}\quad\text{for all }n\in\N.\]

Let's look at the subsequence \((s_{2^{m}})\):

\[s_{2^{1}} = s_{2} = 1 + \frac{1}{2}\]

\[s_{2^{2}} = s_{4} = 1 + \frac{1}{2} + \frac{1}{3} + \frac{1}{4} \geq 1 + \frac{1}{2} + \frac{1}{4} + \frac{1}{4} = 1+\frac{2}{2}\]

\[s_{2^{3}} = s_{8} = 1 + \frac{1}{2} + \left(\frac{1}{3} + \frac{1}{4}\right) + \left(\frac{1}{5} + \frac{1}{6} + \frac{1}{7} + \frac{1}{8}\right)\]

\[\geq 1 + \frac{1}{2} + \left(\frac{1}{4} + \frac{1}{4}\right) + \left(\frac{1}{8} + \frac{1}{8} + \frac{1}{8} + \frac{1}{8}\right) = 1 + \frac{3}{2}\]

\[s_{2^{3}} = s_{8} = 1 + \frac{1}{2} +\frac{1}{3} + \frac{1}{4}+\frac{1}{5} + \frac{1}{6} + \frac{1}{7} + \frac{1}{8}\]

\[\geq 1 + \frac{1}{2}+\frac{1}{4} + \frac{1}{4}+\frac{1}{8} + \frac{1}{8} + \frac{1}{8} + \frac{1}{8}= 1 + \frac{3}{2}\]

Example. Define the sequence \(S = (s_{n})\) by

\[s_{n} = \sum_{k=1}^{n}\frac{1}{k}\quad\text{for all }n\in\N.\]

\[s_{2^{m}} = 1 + \frac{1}{2} + \frac{1}{3} + \frac{1}{4} + \frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}+ \cdots + \frac{1}{2^{m}}\]

\[ = 1 + \left(\frac{1}{2}\right) + \left(\frac{1}{3} + \frac{1}{4}\right) + \left(\frac{1}{5}+\frac{1}{6}+\frac{1}{7}+\frac{1}{8}\right) + \cdots \] \[+ \left(\frac{1}{2^{m-1}+1}+\frac{1}{2^{m-1}+2}+\cdots+\frac{1}{2^{m}}\right)\]

\[ \geq 1 + \left(\frac{1}{2}\right) + \left(\frac{1}{4} + \frac{1}{4}\right) + \left(\frac{1}{8}+\frac{1}{8}+\frac{1}{8}+\frac{1}{8}\right) + \cdots \] \[+ \left(\frac{1}{2^{m}}+\frac{1}{2^{m}}+\cdots+\frac{1}{2^{m}}\right)\]

\[ = 1 + \left(\frac{1}{2}\right) + \left(\frac{1}{2}\right) + \left(\frac{1}{2}\right) + \cdots + \left(\frac{1}{2}\right) \]

\[= 1+\frac{m}{2}\]

Example continued. Hence, given any \(M\in\R\), by the Archimedean Property we can find a natural number \(m>2M-2\).

This implies \(\frac{m}{2}+1>M\), thus

\[s_{2^{m}}\geq 1+\frac{m}{2}>M.\]

This shows that \((s_{n})\) is not bounded, and hence \((s_{n})\) is not Cauchy.

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

 

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded.

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded. By the Bolzano-Weierstrass theorem there is a subsequence \(X' = (x_{n_{k}})\) that converges to some number \(x^{\ast}\).

 

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded. By the Bolzano-Weierstrass theorem there is a subsequence \(X' = (x_{n_{k}})\) that converges to some number \(x^{\ast}\).

Let \(\varepsilon>0\) be given. 

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded. By the Bolzano-Weierstrass theorem there is a subsequence \(X' = (x_{n_{k}})\) that converges to some number \(x^{\ast}\).

Let \(\varepsilon>0\) be given. Since \(X\) is a Cauchy sequence there is a number \(M\in\N\) such that \[|x_{n}-x_{m}|<\frac{\varepsilon}{2}\quad\text{for all }n,m\geq M.\] 

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded. By the Bolzano-Weierstrass theorem there is a subsequence \(X' = (x_{n_{k}})\) that converges to some number \(x^{\ast}\).

Let \(\varepsilon>0\) be given. Since \(X\) is a Cauchy sequence there is a number \(M\in\N\) such that \[|x_{n}-x_{m}|<\frac{\varepsilon}{2}\quad\text{for all }n,m\geq M.\] Since \(X'\) converges to \(x^{\ast}\) there is a number \(K\in\N\) such that \[|x_{n_{k}}-x^{\ast}|<\frac{\varepsilon}{2}\quad\text{ for all }k\geq K.\]

 

Theorem (Cauchy Convergence Criterion). A sequence of real numbers is convergent if and only if it is a Cauchy sequence.

Proof. We have already proved that convergent sequences are Cauchy.

Assume \(X = (x_{n})\) is a Cauchy sequence. We have already seen that \(X\) is bounded. By the Bolzano-Weierstrass theorem there is a subsequence \(X' = (x_{n_{k}})\) that converges to some number \(x^{\ast}\).

Let \(\varepsilon>0\) be given. Since \(X\) is a Cauchy sequence there is a number \(M\in\N\) such that \[|x_{n}-x_{m}|<\frac{\varepsilon}{2}\quad\text{for all }n,m\geq M.\] Since \(X'\) converges to \(x^{\ast}\) there is a number \(K\in\N\) such that \[|x_{n_{k}}-x^{\ast}|<\frac{\varepsilon}{2}\quad\text{ for all }k\geq K.\]

Fix some \(K_{0}\geq K\) such that \(n_{K_{0}}\geq M\). If \(n\geq M\) then we have

\[|x_{n} - x^{\ast}| = |x_{n} - x_{n_{K_{0}}} + x_{n_{K_{0}}} - x^{\ast}| \leq |x_{n} - x_{n_{K_{0}}}|+ |x_{n_{K_{0}}} - x^{\ast}|<\varepsilon.\ \Box\]

Example. Consider the sequence \((x_{n})\) defined by

\[x_{1} = 1,\quad x_{2} = 3,\quad\text{and}\quad x_{n+2}=\frac{1}{2}(x_{n}+x_{n+1})\text{ for all }n\in\N\]

Hence,

\[(x_{n}) = \left(1,3,2,\frac{5}{2},\frac{9}{4},\frac{19}{8},\ldots\right) = \left(\frac{8}{8},\frac{24}{8},\frac{16}{8}\frac{20}{8},\frac{18}{8},\frac{19}{8},\ldots\right)\]

This sequence is neither increasing nor decreasing. Does it converge?

\[(|x_{n}-x_{n+1}|) = \left(2,1,\frac{1}{2},\frac{1}{4},\frac{1}{8},\ldots\right)\]

So, it looks like \(|x_{n}-x_{n+1}|=\frac{1}{2^{n-2}}\) for each \(n\in\N\) (Use induction to prove it!)

By the triangle inequality, for \(m>n\) we have

 

\[|x_{n}-x_{m}| = |x_{n} - x_{n+1} + x_{n+1} - x_{n+2} + x_{n+2} -\cdots +x_{m-1}-x_{m}|\]

\[\leq |x_{n} - x_{n+1}| + |x_{n+1} - x_{n+2}| + \cdots + |x_{m-1}-x_{m}|\]

Example continued. 

\[|x_{n}-x_{m}| = |x_{n} - x_{n+1} + x_{n+1} - x_{n+2} + x_{n+2} -\cdots +x_{m-1}-x_{m}|\]

\[\leq |x_{n} - x_{n+1}| + |x_{n+1} - x_{n+2}| + \cdots + |x_{m-1}-x_{m}|\]

\[= \frac{1}{2^{n-2}} + \frac{1}{2^{n-1}} + \cdots + \frac{1}{2^{m-3}} = \frac{1}{2^{n-2}}\left(1 + \frac{1}{2} + \cdots + \frac{1}{2^{m-n-1}}\right)\]

Where did that last inequality come from? If \(k\in\N\) and \[S = 1+\frac{1}{2} + \frac{1}{4} + \frac{1}{8} + \cdots + \frac{1}{2^{k}}\] then \[2S = 2+1+\frac{1}{2}+\frac{1}{4}+\cdots + \frac{1}{2^{k-1}}\] and subtracting we have \[S = 2S-S = 2-\frac{1}{2^{k}}<2\]

\[< \frac{1}{2^{n-2}}(2) = \frac{1}{2^{n-3}}\]

Example continued. Hence, if we are given any \(\varepsilon>0\), we can choose \(M\in\N\) large enough that

\[\frac{1}{2^{M-3}}<\varepsilon\]

If \(m,n\in\N\) such that \(m,n\geq M\) and \(m>n\), then

\[|x_{n}-x_{m}| < \frac{1}{2^{m-3}}\leq \frac{1}{2^{M-3}}<\varepsilon.\]

This shows that \((x_{n})\) is a Cauchy sequence. By the Cauchy Convergence Criterion the sequence \((x_{n})\) is convergent.

 

What is the limit?

One can prove by induction (you should do it!) that 

\[x_{2n-1} = 1+\frac{1}{3}\left(4-\frac{1}{4^{n-2}}\right)\text{ for all }n\in\N.\]

Hence, \(\lim(x_{2n-1})=1+\frac{4}{3}\), since every subsequence of a convergent sequence converges the limit of the sequence, we conclude \(\lim(x_{n}) = \frac{7}{3}.\)

Practice Problem. Define the sequence \((a_{n})\) where

\[a_{1} = 1\quad\text{and}\quad a_{n} = a_{n-1} + \frac{1}{2^{n}}\quad\text{for all  }n\geq 2\]

Use a proof very similar to the one in the previous example to show that \((a_{n})\) is a Cauchy sequence.

Let \(\varepsilon>0\) be arbitrary. Let \(K\) be large enough that \(\frac{1}{2^{K}}<\varepsilon\), and also take \(K\geq 2\). For any natural number \(n\geq 2\)

\[|x_{n} - x_{n-1}| = \frac{1}{2^{n}}\]

Thus, if \(n,m\geq K\), with \(n\geq m\) we have

\[|x_{n} - x_{m}| = |x_{n} - x_{n-1} + x_{n-1} - x_{n-2} + \cdots + x_{m+1} - x_{m}|\]

\[\leq |x_{n} - x_{n-1}| + |x_{n-1} - x_{n-2}| + \cdots + |x_{m+1} - x_{m}|\]

\[= \frac{1}{2^n} + \frac{1}{2^{n-1}} + \cdots + \frac{1}{2^{m+1}}\]

\[= \frac{1}{2^{m+1}}\left(\frac{1}{2^{n-m-1}}+ \frac{1}{2^{n-m-2}}+\cdots\frac{1}{2} + 1\right)<\frac{1}{2^{m+1}}(2)=\frac{1}{2^{m}}\leq \frac{1}{2^{M}}<\varepsilon\]

Lecture 27

By John Jasper

Lecture 27

  • 256