Let \((x_n)\) be a monotonic sequence.
If \((x_n)\) is bounded then it is convergent
and \[\lim_{n\to\infty} x_n = \sup(x_n, n \in \mathbb{N})\] or \(\inf(x_n, n \in \mathbb{N})\)
\(x_n = \frac{1}{n}\)
\(x_n\) is decreasing and bounded, so it converges to \(\inf(\frac{1}{n}) = 0\)
\(x_n = 1 + \frac{1}{2} + ... + \frac{1}{n}\)
\((x_n)\) is increasing. We will show it is not bounded.
\(x_n = 1 + \frac{1}{2} + (\frac{1}{3} + \frac{1}{4}) + ... \\ \geq 1 + \frac{1}{2} + (\frac{1}{4} + \frac{1}{4}) + ... = 1 + \frac{n}{2}\)
So \(x_n\) is not bounded
Let \(x_n\) be defined
\(x_1 = 1\)
\(x_{n+1} = \frac{1}{4}(2x + 3)\)
First prove that \(x_n\) is increasing. (by induction)
\(x_1 < x_2\)
\(\frac{1}{4}(2x_n + 3) \leq \frac{1}{4}(2x_{n+1} + 3)\)
so \(x_n \leq x_{n+1}\)
Show bounded (by induction)
\(x_1 \leq 2\)
Assume \(x_n \leq 2\), then
\(x_{n+1} = \frac{1}{4}(2x_n + 3) \leq \frac{7}{4} \leq 2\)
so \(x_n\) is bounded.
Now to find the limit. \(x_n = x_{n+1}\)
so \(x = \frac{1}{4}(2x + 3) \Rightarrow x = \frac{3}{2}\)
subsequence
Let \((y_n)\) be a sequence.
If the sequence \((x_n) \subseteq (y_n)\)
Then \((x_n)\) is a subsequence of \((y_n)\)
\((2,4,6,8,...)\) is a subsequence of \((1,2,3,4,...)\)
\(x_n = (-1)^n\) \((-1,1,-1,1,...)\)
\(x_n = 1\) \((1,1,1,1,...)\)
\(x_n = -1\) \((-1,-1,-1,-1,...)\)
Let \((x_n)\) be a convergent sequence
Then all subsequences are also convergent and the limits are the same.
\(x_n = \frac{1}{n^4}\)
Since \(n^4 \in \mathbb{N}\), \(x_n\) is a subsequence of \(\frac{1}{n}\).
So \(x_n\) must also converge to 0.
\(x_n = (-1)^n\)
\[\lim_{n\to\infty} x_{2n} = 1\]
\[\lim_{n\to\infty} x_{2n+1}= -1\]
So \(x_n\) is divergent
Similarly, if \((x_n)\) is a sequence and \((y_n)\) is a divergent subsequence
then \((x_n)\) is divergent
Let \((x_n)\) be a sequence.
There exists a subsequence \((x_m)\) which is monotonic
Let \((x_n)\) be a bounded sequence
Then there exists a subsequence \((x_m)\) that is convergent
\(x_n = (-1)^n\)
\(x_{2n} \Rightarrow 1\)
\(x_{2n+1} \Rightarrow -1\)
Any subsequence of \((x_n) = \sin(n)\) can converge to any \(x \in [-1,1]\)
cauchy sequence
Let \((x_n)\) be a sequence.
\((x_n)\) is a cauchy sequence if \(\forall \varepsilon > 0\) \(\exists K(\varepsilon) \geq 0\)
\(\forall n,m \geq K(\varepsilon), | x_n - x_m | \leq \varepsilon\)
Let \((x_n)\) be a convergent sequence.
Then it is a Cauchy sequence
The inverse is not true
Not a Cauchy sequence:
\((x_n) = \sqrt{n}\)
\((x_n)\) is Cauchy, but does not converge
Let \((x_n)\) be a bounded Cauchy sequence
Then \((x_n)\) is convergent
\(x_n = 1 + \frac{1}{2} + ... + \frac{1}{n}\)
\(x_{2n} - x_n = \frac{1}{n+1} + \frac{1}{n+2} + ... + \frac{1}{2n} \geq \frac{1}{2n} + ... + \geq \frac{1}{2n}\)
So \(x_{2n} - x_n\) does not converge to \(0\) therefore \((x_n)\) is not a Cauchy sequence
contracting sequence
Let \((x_n)\) be a sequence
\((x_n)\) is contracting if
\[\exists C > 0, \forall n \in \mathbb{N}, | x_{n+1} - x_n | \leq C | x_n - x_{n-1} |\]
Any contracting sequence is convergent
xn = n
Let \(M \geq 0\)
for \(n \geq M, x_n \geq M\)
So \((x_n)\) goes to infinity\(x_n = n^2 - n = n(n-1) \geq (n-1)^2\)
Let \((x_n)\) be increasing and not bounded. Then \[\lim_{n \to \infty} x_n = \infty\]
Let \((x_n)\) be decreasing and not bounded. Then \[\lim_{n \to -\infty} x_n = -\infty\]
Let \((x_n),(y_n)\) be sequences s.t. \((x_n) \geq (y_n)\)
\[\lim_{n \to \infty} (y_n) = \infty\] implies
\[\lim_{n \to \infty} (y_n) = \infty\]
series
A sequence generated by the sum of the first \(n\) terms of another sequence
The series \((s_n)\) generated by \((x_n)\) lookks like:
\(s_1 = x_1\)
\(s_2 = x_1 + x_2\)
…
\(s_n = x_1 + ... + x_n\)
limit of series
\[\lim_{n \to \infty} s_n = \sum_{k=1}^{\infty} x_k\]
Let \((x_n) = \frac{1}{n(n+1)}\)
Then
\[ \begin{aligned} (s_n) & = \sum_{k=1}^{\infty} \frac{1}{k(k+1)} \\ & = \sum_{k=1}^{\infty} \frac{1}{k} - \sum_{k=1}^{\infty} \frac{1}{k+1} \\ & = \sum_{k=1}^{\infty} \frac{1}{k} - \sum_{k=2}^{\infty} \frac{1}{k} \\ & = 1 - \frac{1}{n+1} \end{aligned} \]
Let \((x_n) = r^{n-1}\)
then
\[ \begin{aligned} s_n & = \sum_{k=1}^{\infty} r^{n-1} \\ & = \frac{1-r^n}{1-r} \end{aligned} \]
case \(|r| < 1\)
\[\lim_{n \to \infty} r^n = 0\]
So \[\sum_{k=1}^{\infty} r^{n-1} = \frac{1}{1-n}\]
and \((s_n)\) is convergent
case \(|r| = 1\)
\[\sum_{k=1}^{\infty} x_k = \infty\]
\(s_n = \frac{1}{2} (1 - (-1^n))\)
So \((s_n)\) is divergent
Let \(x_n = \frac{1}{2^n}\)
then \[\lim_{n \to \infty} s_n = \sum_{k=1}^{\infty} \frac{1}{2^k} = 2\]
Let \((s_n)\) be a series generated by the sequence \((x_n)\)
If \((s_n)\) converges, then \((x_n)\) also converges
Convergence of \((x_n)\) does not imply convergence of \((s_n)\)
ex. harmonic series = \(x_n = \frac{1}{n}\)
If \((x_n)\) is a positive sequence
then \[0 < \lim_{n \to \infty} s_n \leq \infty\]
Let \((x_n)\) and \((y_n)\) be sequences s.t. \(0 \leq x_n \leq y_n\)
Let \((s_n)\) and \((t_n)\) be sequences of \((x_n)\) and \((y_n)\) respectively
If \((t_n)\) is convergent, then \((s_n)\) is convergent
If \((s_n)\) is divergent, then \((t_n)\) is divergent
Let \(x_n = \frac{1}{n^r}, r \geq 0\)
then
\[ \begin{aligned} s_n = \sum_{k=1}^{\infty} \frac{1}{k^r} \ & = 1 + \frac{1}{2^r} + ... \end{aligned} \]
case \(0 \leq r \leq 1\)
\(\frac{1}{n^r} \geq \frac{1}{n}\)
So \(s_n\) is divergent
case \(r > 1\)
For \(n \geq 2\) (by bernoulli inequality)
\(\frac{1}{nr} \leq \frac{1}{r-1} (\frac{1}{(n-1)^{r-1}} - \frac{1}{n^{r-1}})\)
Let \(y_n = \frac{1}{r-1} (\frac{1}{(n-1)^r} - \frac{1}{n^{r-1}})\)
Then \[\sum_{k=1}^{\infty} y_k = \frac{1}{r-1}(1 - \frac{1}{n^{r-1}}) < \infty\]
So \[\sum_{k=1}^{\infty} x_k < \infty\]
We conclude \[\sum_{k=1}^{\infty} \frac{1}{n^r} < \infty\] if and and only if \(r > 1\)
cluster point
Let \(A \subseteq \mathbb{R}\) be a non empty set.
Then \(c \in \mathbb{R}\) is a cluster point of A if
\(\exists a \in A\), \(a \neq c\) s.t. \(\forall \varepsilon > 0\),
\(|a - c| < \varepsilon\)
\(c\) is a cluster point of the set \(A\) if the values of \(A\) get arbitrarily close to \(c\)
Let \(A = \{ \frac{1}{n}, n \in \mathbb{N}\}\)
0 is the only cluster point for the set \(A\)
Let \(A = \{n | n \in \mathbb{N}\}\)
There are no cluster points in the set. We can always find a smaller \(\varepsilon\) which excludes \(c\)
Let \(A = \mathbb{Q} = \{\frac{x}{y} | x,y \in \mathbb{N}\}\)
The set of cluster points is \(\mathbb{R}\)
This is because \(\mathbb{Q}\) is dense in \(\mathbb{R}\)
function
Let \(A \subset \mathbb{R}\) be a non empty set.
Then the mapping \(f: A \rightarrow \mathbb{R}\) is a function
limit
\(f\) admits a limit \(L \in \mathbb{R}\) if:
\(\forall \varepsilon > 0\), \(\exists \zeta > 0\),
s.t. \(|x - a| \leq \zeta\) implies \(|f(x) - L| \leq \varepsilon\)
Or, as \(x\) converges to \(a\), \(f(x)\) converges to \(L\)
Let \(f(x) = x\)
Let \(a \in \mathbb{R}\), \(\varepsilon > 0\)
if \(|x - a| \leq \varepsilon\),
then \(|f(x) - a| \leq \varepsilon\)
Uniqueness of Limit
Let \(A \subseteq \mathbb{R}\) be a nonempty set
Let \(f: A \to \mathbb{R}\) and let \(c \in \mathbb{R}\) be a cluster point of \(A\)
Assume \(f\) has a limit \(L\) at \(c\) and also has limit \(L'\) at \(c\)
Then \(L = L'\)
and \(L = \lim_{x \to c} f(x)\) is called the limit of \(f\) at \(c\)
find the limit of \(f(x) = a, a \in \mathbb{R}\)
Let \(\varepsilon > 0\)
For \(x \in \mathbb{R}\), s.t. \(|x - 1| \leq 1\)
we have \(|f(x) - a| < \varepsilon\)
Let \(f(x) = x\). Show \(\lim_{x \to 1} f(x) = 1\)
Let \(\varepsilon > 0\)
for \(x \in \mathbb{R}\) s.t. \(|x - 1| < \varepsilon\) we have \(|f(x) - 1| = |x - 1| < \varepsilon\)
Let \(f(x) = x^2\) Show \(\lim_{x \to a} f(x) = a^2\)
Let \(\varepsilon > 0\)
\(|x^2 - a^2| = |x-a||x+a|\)
If \(|x - a| \leq 1\), \(|x| \leq 1 + |a|\),
so \(|x + a| \leq 1 + 2 |a|\)
\(|x^2 - a^2| \leq |x-a|(1 + 2 |a|)\)
Let \(\zeta = \min(1, \frac{\varepsilon}{1 + 2 |a|})\)
For \(x \in \mathbb{R}\) s.t. \(|x - a| \leq \zeta\),
we have \(|x^2 - a^2| < \varepsilon\)
Let \(f(x) = \frac{1}{x}, x > 0\). Prove \(\lim_{x \to a} \frac{1}{x} = \frac{1}{a}\)
\(|\frac{1}{x} - \frac{1}{a}| = \frac{|a - x|}{|a||x|}\)
For \(|x - a| < \frac{a}{2}, x > \frac{a}{2}\)
we have \(|\frac{1}{x} - \frac{1}{a}| \leq \frac{2}{a^2} |a - x|\)
Let \(\zeta = \min(\frac{a}{2}, \frac{\varepsilon a^2}{2})\)
if \(|x - a| \leq \zeta\)
then \(|\frac{1}{x} - \frac{1}{a}| \leq \varepsilon\)
Let \(A \subset \mathbb{R}\) be a nonempty set and \(f: A \rightarrow \mathbb{R}\)
Let c be a cluster point of A. Then the conditions are equivalent:
For every sequence \((x_n)\) in A where \(x_n \neq c\) and \(\lim_{n \to \infty} x_n = c\),
then \(\lim_{n \to \infty} f(x_n) = c\)
Let \(A \subset \mathbb{R}\) be nonempty, \(f:A \rightarrow \mathbb{R}\), and \(c \in \mathbb{R}\) be a cluster point of A
Assume that you can find a sequence \((x_n)\) in A, \(x_n \neq c\) and \(x_n \rightarrow c\), but \(f(x_n)\) is not convergent
Then \(\lim_{x \ to c} f(x)\) does not exist
Let \(f(x) = \frac{1}{x}, x > 0\)
Let $xn = . We have $xn → 0, \(f(x_n)\) is convergent so
\(\lim_{x \to 0} f(x)\) does not exist
Show the function does not converge
\[ f(x) = \begin{cases} 1 & x \in \mathbb{Q} \\ 0 & \text{else} \end{cases} \]
Let \((x_n) = \frac{1}{n}\). \(\lim_{n \to \infty} f(x_n) = 1\)
Let \((y_n) = \frac{\sqrt{2}}{n}\). \(\lim_{n \to \infty} f(x_n) = 0\)
Since the subsequences don't converge to the same value, \(f(x)\) does not converge
Let \(A \subset \mathbb{R}\) be a nonempty set. Let \(c \in \mathbb{R}\) be a cluster point of \(A\). Let \(f: A \longrightarrow \mathbb{R}\)
Then \(f\) is bounded in a neighborhood of c if there exists \(M > 0\) and \(\zeta > 0\)
s.t. if \(x\in[c - \zeta, c + \zeta] \cap A\)
\(|f(x) \leq M\)
Let \(f(x) = x^2\). For \(x \in [-1,1]\), \(f(x) \leq 1\) so \(f\) is bounded in the neighborhood of 0
Let \(f(x) = \frac{1}{x}\)
Let \(f: A \longrightarrow \mathbb{R}\) Let \(c\) be a cluster point of \(A\).
If \(\lim_{x \to c} f(x)\) exists, then \(f\) is a bounded in a neighborhood of \(c\)
Let \(f,g: A \longrightarrow \mathbb{R}\)
if \(\forall x \in A, g(x) \neq 0\)
\(\frac{f}{g}(x) = \frac{f(x)}{g(x)}\)
Let \(f,g: A \longrightarrow \mathbb{R}\)
Let \(c \in \mathbb{R}\) be a cluster point of \(A\)
If \(\lim_{x \to c} f(x)\) and \(\lim_{x \to c} g(x)\) exist, then
\(\lim_{x \to c} f(x) + \lim_{x \to c} g(x) = \lim_{x \to c} f(x) + g(x)\)
Let \(f,g: A \to \mathbb{R}\). Let \(c \in \mathbb{R}\) be a cluster point of \(A\).
If \(\lim_{x \to c} f(x)\) and lim x → c g(x)$ exist, then
\(\lim_{x \to c} f(x)g(x) = \lim_{x \to c} f(x) \lim_{x \to c} g(x)\)
Find \(\lim_{x \to 1} \frac{x^2 + 1}{x}\)
\(\lim_{x \to 1} \frac{x^2 + 1}{x} = \frac{\lim_{x \to 1} x^2 + 1}{\lim_{x \to 1} x} = 2\)
Let \(f(x) = x + \mathbb{1}_{\mathbb{Q}}\)
The limit of \(x\) exists, but not \(\mathbb{1}_{\mathbb{Q}}\), so the limit of \(f(x)\) does not exist
Let \(f(x) = x \mathbb{1}_{\mathbb{Q}}\)
Since \(\mathbb{1}_{\mathbb{Q}}\) is bounded and \(x\) is convergent, \(f(x)\) is convergent
Let \(f,g,h: A \to \mathbb{R}\). Let \(c \in \mathbb{R}\) be a cluster point of \(A\).
Assume \(f \leq g \leq h\) for all \(x\).
If \(\lim_{x \to c} f(x) = \lim_{x \to c} h(x) = L\), then \(\lim_{x \to c} g(x) = L\)
M > 0. prop div if xn > M for all n > k
Let \(f: A \to \mathbb{R}\) Let \(c \in \mathbb{R}\) be a cluster point of \(A\)
We say that \(\lim_{x \to c} f(x) = \infty\)
if for \(\forall M \geq 0\), \(\exists \zeta > 0\), s.t. for \(x \in [c - \zeta, c + \zeta] \cap A, x \neq c\)
\(f(x) \geq M\)
\(f(x) = \frac{1}{x}, x > 0\)
\(\lim_{x \to 0} f(x) = \infty\)
Let \(f(x) = \frac{1}{x}, x < 0\)
\(\lim_{x \to 0} f(x) = -\infty\)
changing A can change the limit
Let \(f: A \to \mathbb{R}\). Let \(c \in \mathbb{R}\) be a cluster point of \(A\)
then \(\lim_{x \to c} f(x) = \infty\) iff
for every sequence \((x_n) \in A\) converging to \(c\), \(x_n \neq c\)
\(\lim_{n \to \infty} f(x_n) = \infty\)
Continuous at point
Let \(\varepsilon > 0\).
\(f\) is continuous at \(c\) if there exists \(\zeta > 0\) such that
\(\forall x \in A, |x - c| \leq \zeta\) implies \(|f(x) - f(c)| \leq \varepsilon\)
If \(c \in A\) is a cluster point of A,
then \(f\) is a continuous function at \(c\) iff \(\lim_{x \to c} f(x) = f(c)\)
If \(c \in A\) is not a cluster point, then \(f\) is automatically continuous at \(c\)
Let \(f(x) = x, x \in \mathbb{Z}\).
Then \(f\) is continuous at \(x\)
Let
\[ f(x) = \begin{cases} 1 & x \geq 0 \\ 0 & x < 0 \end{cases} \]
Then \(f\) is not continuous at 0
Let
\[ f(x) = \begin{cases} 1 & x \in \mathbb{Q} \\ 0 & x \notin \mathbb{Q} \end{cases} \]
f(x) is nowhere continuous
Let
\[ f(x) = \begin{cases} 1 & x \in \mathbb{R} \\ 0 & x \notin \mathbb{R} \end{cases} \]
from the squeeze theorem, \(f\) is continuous at 0
but if \(x \neq 0\), then the limit does not exist and \(f\) is not continuous
Continuity of points
Let \(A \rightarrow \mathbb{R}\)
Let \(c \in A\), then \(f\) is continuous at \(c\) iff
for every \((x_n)\) s.t. \(\lim_{n\to \infty} x_n = c\), we have \(\lim_{n\to \infty} f(x_n) = f(c)\)
\(f\) is not continuous at \(c\), iff there exists \((x_n)\) that converges to c
where \(f(x_n)\) does not converge to \(f(c)\)
Continuity of functions
Let \(f:A \to \mathbb{R}\)
Let \(B \subset A\)
We say \(f\) is continuous on \(B\) if \(f(x)\) is continuous for every \(x \in B\)
Let \(f(x) = 1, x > 0\)
\(f\) is continuous on \((0, \infty)\)
Let \(f:A \to \mathbb{R}\) and \(c \in \mathbb{R}\) be a cluster point of \(A\).
Assume \(c \notin A\), but \(\lim_{x \to c} f(x)\) exists. Then we can define
\(f'(x) = f(x), x in A, \lim_{x \to c}f(x), x = c\)
f' fills in holes
Let \(f,g:A \to \mathbb{R}\) be continuous functions at \(c\). Let \(c \in A\)
Then \(f+g\), \(fg\), and \(\frac{f}{g}\) (\(g \neq 0\)) are continuous at \(c\)
Let \(f\) be any polynomial in \(\mathbb{R}\)
Since \(f\) is a sum of products of \(x\) and \(x\) is continuous on \(\mathbb{R}\), \(f\) is also continuous on \(\mathbb{R}\)
Let \(f:A \to \mathbb{R},g:B \to \mathbb{R}\)
Assume for \(x \in B\), \(g(x) \in A\)
We define \((f \circ g)(x) = f(g(x))\)
If \(g\) is continuous at \(c\) and \(f\) is continuous at \(g(c)\), then \(f \circ g\) is continuous at \(c\)
Let \(A,B \subseteq \mathbb{R}\). Let \(f\) be continuous on \(A\), and \(g\) be continuous on \(B\)
If \(f(A) \subseteq B\), then \(g \circ f\) is continuous on \(A\)
closed set/function
interval is a continuous subset of R
f has a supremum on a bounded set uses squeeze theorem
Assume that \(f\) is not bounded on \(I\). We can assume \(f\) has no upper bound.
\(n\) is not an upper bound for \(f\), so we can find \((x_n) \in I\) s.t. \(f(x_n) \geq n\)
\((x_n)\) is bounded, therefore it admits a convergent subsequence \((x_m)\)
Let \(x^* = \lim x_m\)
Since I is closed, \(x^* \in I\)
Since \(f\) is continuous at \(x^* \lim f(x_m) = \lim f(x^*)\)
This contradicts \(f(x_n) \geq n\)
So \(f\) admits a supremum on \(I\)
The proof is similar for infimum
The supremum/infimum of a function of closed set is also a min/max
Let \(I = [a,b]\). Let \(f:I \to \mathbb{R}\) be a continuous function.
Let \(M = \text{sup}{f(x), x \in I}\), \(S = \inf {f(x), x \in I}\)
\(\exists x_S, x_M \in I\) s.t. \(f(x_S) = S, f(x_M) = M\)
Let \(f(x) = x^2, I = [-1, 1]\)
\(\sup{F} = 1 = f(1)\)
\(\inf{F} = 0 = f(0)\)
Let \(f(x) = x, I = (0,1)\)
\(\inf f = 0\), but there is no \(x \in I\) such that \(f(x) = 0\)
Intermediate Value Theorem
Let \(I\) be an interval and \(f:I \to \mathbb{R}\) be continuous
Let \(a,b,c \in I\) s.t. \(a \leq b\) \(f(a) \leq f(b)\)
Let \(\lambda \in [f(a), f(b)]\). We can always find \(c \in [a,b]\)
s.t. \(f(c) = \lambda\)
Zero crossing corollary
Let \(f:[a,b] \to \mathbb{R}\) be continuous
If \(f(a) < 0, f(b) > 0\) then \(\exists c\) s.t. \(f(c) = 0\)
Let \(f:\mathbb{R} \to \mathbb{R}\) be continuous. Let \(I \in \mathbb{R}\) be an interval.
Then \(f(I) = {f(x) | x \in I}\) is an interval
\(f\) is uniformly continuous if
for all \(\varepsilon\) there exists \(\delta\) such that
\(|x_1 - x_2| \leq \delta\) implies \(|f(x_1) - f(x_2)| \leq \varepsilon\)
Lipschitz
Let \(f:A \to \mathbb{R}\)
\(f\) is lipschitz if, there exists \(K \geq 0\) such that
for all \(x, y \in A\)
\(|f(x) - f(y)| \leq K|x - y|\)
Let \(f(x) = x\).
\(|f(x) - f(y)| \leq 1 * |x - y|\), so \(f\) is lipschitz
Let \(f(x) = \sqrt{x}\)
\(|f(x) - f(y)| = |\sqrt{x} - \sqrt{y}| = \frac{1}{\sqrt{x} + \sqrt{y}} |x - y|\)
so \(f\) is not lipschitz on \([0, \infty)\).
If we let \(x,y \in [a, \infty)\), then
\(|f(x) - f(y)| \leq \frac{1}{2\sqrt{a}} |x - y|\)
so \(f\) is lipschitz on \([a, \infty)\)
Let \(f(x) = x^2\)
\(|f(x) - f(y)| = |x^2 - y^2| = |x - y||x + y|\).
\(f\) is not lipschitz on \(\mathbb{R}\)
but it is for any bounded interval, since some \(K\) will be greater than \(|x - y|\)
Any lipschitz function is uniformly continuous
The converse is not true, as \(f(x) = \sqrt{x}\) is uniformly continuous, but not lipschitz.
ie. \(\sqrt{x}\) is continuous on a bounded interval \([0,1]\), but the derivative is not bounded (\(\lim_{x \to 0} f'(x) = \infty\))
The derivative of Lipschitz functions are bounded by K
*Increasing and decreasing functions
Let \(f:A \to \mathbb{R}\), and \(x,y \in A\), such that \(x \leq y\)
then \(f\) is increasing if \(f(x) \leq f(y)\)
the definition is similar for strictly increasing/decreasing
Let \(f(x) = x\)
\(f\) is strictly increasing
Let \(f(x) = x^2\)
\(f\) is strictly increasing on the domain \(x \geq 0\)
The left and right limits exist for all points in monotonic functions.
Let \(f:I \to \mathbb{R}\) be monotonic.
Let \(c \in I\) which is not an endpoint of \(I\).
Then \(\lim_{x \to +c} f(x)\) and \(\lim_{x \to -c} f(x)\) exist
Inverse The function \(g:f(A) \to A\) such that \(g(f(a)) = a\) for \(a \in A\)
A function \(f\) has an inverse if \(f\) is injective. (\(f\) uniquely maps values in its domain)
ie. \(x \neq y\) implies \(f(x) \neq f(y)\)
Let \(f:A \to \mathbb{R}\) be a strictly increasing function.
Then \(f\) is injective and \(g:f(A) \to A\) is strictly increasing.
Let \(f\) be a strictly increasing and continuous function. Then \(f^{-1}\) is strictly increasing and continuous.
Differentiability
We say that a function \(f:I_1 \to \mathbb{R}\) is differentiable at \(c \in I\) if
\[\lim_{x \to c} \dfrac{f(x) - f(c)}{x - c}\]
exists. We call this \(f'(c)\)
Let \(f(x) = c\) for \(c \in \mathbb{R}\)
\(\frac{f(x) - f(c)}{x - c} = 0\)
so \(f'(x) = 0\)
Let \(f(x) = x\)
\(\frac{f(x) - f(c)}{x - c} = \frac{x - c}{x - c} = 1\)
so \(\lim_{x \to c} f'(x) = 1\)
Let \(f(x) = x^2\)
\(\frac{f(x) - f(c)}{x - c} = \frac{x^2 - c^2}{x - c} = \frac{(x - c)(x + c)}{x - c} = x + c\)
so \(\lim_{x \to c} x + c = 2c\)
Let \(f(x) = x^3\)
\(\frac{f(x) - f(c)}{x - c} = \frac{x^3 - c^3}{x - c} = \frac{(x - c)(x^2 + xc + c^2)}{x - c} = x^2 + xc + c^2\)
so \(\lim_{x \to c} x^2 + xc + c^2 = 3c^2\)
If a function is differentiable at c, it is continuous at c
Sum Rule
The derivative of a sum of functions is the sum of the derivative of the functions
\((f + g)'(x) = f'(x) = g'(x)\)
Product Rule
\((fg)'(c) = f'(c)g(c) = f(c)g'(c)\)
Quotient Rule
Let \(f\) be a differentiable function which is not zero on a neighborhood around \(c\)
Then \((\frac{1}{f})'(c) = \frac{f'(c)}{f(c)^2}\)
Let \(f,g\) be differentiable functions where \(g\) is not zero on a neighborhood around \(c\)
then \(\left(\frac{f}{g}\right)'(c) = \frac{f'g(c) - fg'(c)}{g(c)^2}\)
The Carathéoday Lemma is needed to prove the next theorem.
Carathéoday Lemma
Let \(f:I \to \mathbb{R}\) be continuous.
\(f\) is differentiable at \(c \in I\) iff,
there exists a continuous \(\phi:I \to \mathbb{R}\) such that
\(f(x) - f(c) = \phi(x)(x - c)\) for all \(x \in I\)
Chain rule
Let \(f:I \to \mathbb{R}\) and \(g:f(I) \to \mathbb{R}\) be continuous.
then \(g(f(c))' = g'(f(c))f'(c)\)
Derivative of inverse
Let \(f:I \to \mathbb{R}\) and \(f^{-1}:f(I) \to I\) be continuous.
then \(f'^{-1}(c) = \frac{1}{f'(g(f(c)))}\)
Relative Extrema
Let \(f: I \to \mathbb{R}\) be continuous.
We say \(f\) has a relative minimum at \(x_0 \in I\) if there exists \(\zeta\) s.t.
for \(x \in [x_0 - \zeta, x_0 + \zeta ] \cap I\)
\(f(x) \geq f(x_0)\)
Note that by the above definition, constant functions are maximum and minimums everywhere.
Derivative at relative extrema
Let \(f: I \to \mathbb{R}\), \(x_0 \in I\), where \(x_0\) is not a boundary point
Assume \(x_0\) is a relative extrema and \(f\) is differentiable at \(x_0\)
Then \(f'(x_0) = 0\)
Note that the converse is not true. \(\frac{d}{dx}\left[ x^3 \right]_{x = 0} = 0\), but \(x = 0\) is not an extremum
Let \(f(x) = |x|\)
There is a relative minimum at 0, but the function is not differentiable
Rolle's Theorem
Let \(f:I \to \mathbb{R}\) be a differentiable function.
If \(f(a) = f(b)\), where \(a < b\)
then \(f'(c) = 0\) for some \(c \in (a,b)\)
if abs(x) is not differentiable, then what is the integral of step function
Mean Value Theorem
Let \(f:I \to \mathbb{R}\) be differentiable over \([a,b]\), where \(a < b\)
then there is a point \(c \in (a,b)\) such that \(f'(c) = \frac{f(b) - f(a)}{b - a}\)
Derivative is zero implies constant function
Let \(f:I \to \mathbb{R}\) be differentiable such that \(f'(x) = 0\) for all \(x\)
Then \(f(x)\) is constant on \(I\)
Positive derivatives implies increasing function
Let \(f:I \to \mathbb{R}\) be differentiable.
\(f\) is increasing iff \(f'(x) \geq 0\)
Let \(f:I \to \mathbb{R}\)
If \(f'(c) = 0\) and \(f'(x) \geq 0\) for \(x \in (c - \delta, c]\)
and \(f'(x) \leq 0\) for \(x \in [c,c + \delta)\)
then \(c\) is a relative minimum.
Indeterminate limit
Let \(f,g:(a,b) \to \mathbb{R}\) be continuous.
If \(\lim_{x \to a+)} f(x) = 0\) and \(\lim_{x \to a+)} g(x) = 0\)
Then we say \[\lim_{x \to a+} \frac{f(x)}{g(x)}\]
is indeterminate
If it exists, evaluate \(\lim_{x \to 0} \frac{sin(x)}{x}\)
\(\frac{sin(x)}{x} = \frac{sin(x) - sin(0)}{x - 0}\)
Since \(\sin(x)\) differentiable at 0, \(\lim_{x \to 0} \frac{\sin(x)}{x} = \cos(0) = 1\)
If it exists, evaluate \(\lim_{x \to 0} \frac{x}{x^2}\)
\(\lim_{x \to 0} \frac{x}{x^2} = \infty\)
so it is indeterminate
L'Hôpital's Rule #1
Let \(f,g:I \to \mathbb{R}\) be continuous.
Assume \(f(a) = g(a) = 0\) and that \(f,g\) are differentiable at \(a\)
and \(g'(a) \neq 0\)
we have \[\lim_{x \to a} \frac{f(x)}{g(x)} = \frac{f'(a)}{g'(a)}\]
If it exists, compute \(\lim_{x \to 0} \frac{1 - \cos(x)}{x}\)
Let \(f(x) = 1 - \cos(x)\), \(g(x) = x\)
Then \(f'(0) = 0\), \(g'(0) = 1\)
so by L'Hôpital's rule, \(\lim_{x \to 0} \frac{1 - \cos(x)}{x}\)
If it exists \(\lim_{x \to 0} \frac{1 - \cos(x)}{x^2}\)
We apply L'Hôpital's rule twice:
\(f(x) = 1 - \cos(x)\), \(f'(x) = \sin(x)\), \(f''(x) = \cos(x)\)
\(g(x) = x^2\), \(g'(x) = 2x\), \(g''(x) = 2\)
so \(\lim_{x \to 0} \frac{1 - \cos(x)}{x^2} = \frac{f''(0)}{g''(0)} = \frac{1}{2}\)
L'Hôpital's Rule #2
Let \(f,g:I \to \mathbb{R}\) be continuous ad differentiable functions.
Assume \(\lim_{x \to a+} f(x),g(x) = 0\) and \(g'(x) \neq 0\) for \(x \in (a,b)\)
a) If \(\lim_{x \to a+} \frac{f'(x)}{g'(x)} = L\), then \(\lim_{x \to a+} \frac{f(x)}{g(x)} = L\) b) If \(\lim_{x \to a+} \frac{f'(x)}{g'(x)} = \infty\), then \(\lim_{x \to a+} \frac{f(x)}{g(x)} = \infty\)
This does not require that \(f,g\) be differentiable at \(a\)
To prove the above, we need the Cauchy Mean Value Theorem
Cauchy Mean Value Theorem
Let \(f,g:I \to \mathbb{R} be differentiable\)
Let \(a,b \in I, a < b\)
Assume \(g'(x) \neq 0\), for \(x \in (a,b)\)
You can find \(c \in (a,b)\) such that
\[\frac{f(b) - f(a)}{g(b) - g(a)} = \frac{f'(c)}{g'(c)}\]
Evaluate \(\lim_{x \to 0} \frac{e^x - 1 - x}{x^2}\)
\(f(x) = e^x - 1 - x\), \(f'(x) = e^x - 1\), \(f''(x) = e^x\)
\(g(x) = x^2\), \(g'(x) = 2x\), \(g''(x) = 2\)
\(\lim_{x \to 0} \frac{f(x)}{g(x)} = \lim_{x \to 0} \frac{f'(x)}{g'(x)} = \lim_{x \to 0} \frac{f''(x)}{g''(x)} = \frac{1}{2}\)
Partition
Let \(I=[a,b]\) A partition of \(I\) is a sequence \(x_0,...,x_n\)
such that \(a = x_0 < ... < x_n = b\)
Tagged partition
A tagged partition is a partition where a tag \(t_i\) is defined for every interval.
where \(t_i \in [x_{i-1}, x_i]\)
Mesh
The mesh of a partition is defined as \(||\cal{P}|| = max(x_i - x_{i - 1} , 1 \leq i \leq n)\)
\[ \lim_{\Delta x \to 0} \sum_{n = 0}^{\frac{b - a}{\Delta x}} \Delta x (f(a + \Delta x n))\]
Riemann Sum
\(S(f, \bar{\cal{P}}) = \sum_{i = 1}{n} f(t_i)(x_i - x_{i-1})\)
Riemann Integrable We say a function \(f\) is Riemann integrable f \(\forall \varepsilon > 0\), you can find \(\delta > 0\) such that
for every tagged partition \(\bar{\cal{P}}\) with mesh \(||\bar{\cal{P}}|| \leq \delta\)
you have \(|S(f,\bar{\cal{P}}) - L| \leq \varepsilon\) and \(L\) is called the Riemann integral, where \(L = \int_a^b f(x) dx\)
Let \(f(x) = 0, x \in [0,1]\)
\(S(f,\cal{\bar{P}}) = 0\) for any tagged partition.
Therefore \(\int_a^b f(x) dx = 0\)
Let \(f(x) = 1, x \in \mathbb{R}\)
\[ \begin{aligned} S(f,\bar{\cal{P}}) & = \sum_{i = 1}^n f(t_i)(x_i - x_{i-1}) \\ & = \sum_{i=1}^n (x_i - x_{i - 1}) \\ & = b - a \end{aligned} \]
So \(\int_a^b f(x) dx = b - a\)
Properties of integrals Let \(f,g \in \cal{R}[a,b]\)
If \(f:I \to \mathbb{R}\) is continuous, then \(f\) is Riemann integrable.
The converse is not true
If \(f\) is a monotonic function, then it is Riemann integrable