Convergence: \(\lim_{n \to \infty} a_n = L\) means for every \(\epsilon > 0\), there exists \(N\) such that \(|a_n - L| < \epsilon\) for all \(n > N\)
Formal definition: \(\forall \epsilon > 0, \exists N \in \mathbb{N}: n > N \Rightarrow |a_n - L| < \epsilon\)
- Guess the limit L by analyzing the sequence behavior
- Start with the inequality \(|a_n - L| < \epsilon\)
- Manipulate the inequality to find N in terms of ε
- Verify that the choice of N works for the definition
\(\lim_{n \to \infty} \frac{2n + 3}{n + 1}\)
Divide numerator and denominator by n:
\(\lim_{n \to \infty} \frac{2 + \frac{3}{n}}{1 + \frac{1}{n}} = \frac{2 + 0}{1 + 0} = 2\)
So we conjecture that \(L = 2\).
We need to show: \(\left|\frac{2n + 3}{n + 1} - 2\right| < \epsilon\)
\(\left|\frac{2n + 3}{n + 1} - 2\right| = \left|\frac{2n + 3 - 2(n + 1)}{n + 1}\right| = \left|\frac{2n + 3 - 2n - 2}{n + 1}\right| = \left|\frac{1}{n + 1}\right|\)
Since \(n > 0\), we have \(n + 1 > 0\), so \(\left|\frac{1}{n + 1}\right| = \frac{1}{n + 1}\)
We need: \(\frac{1}{n + 1} < \epsilon\)
This is equivalent to: \(n + 1 > \frac{1}{\epsilon}\)
Which gives: \(n > \frac{1}{\epsilon} - 1\)
Let \(N = \left\lceil \frac{1}{\epsilon} - 1 \right\rceil\), where \(\lceil x \rceil\) is the ceiling function.
For any \(\epsilon > 0\), let \(N = \left\lceil \frac{1}{\epsilon} - 1 \right\rceil\).
Then for all \(n > N\):
\(n > \frac{1}{\epsilon} - 1\), which implies \(n + 1 > \frac{1}{\epsilon}\)
Therefore: \(\frac{1}{n + 1} < \epsilon\), which means \(\left|\frac{2n + 3}{n + 1} - 2\right| < \epsilon\)
Convergent sequence
The sequence \(a_n = \frac{2n + 3}{n + 1}\) converges to 2.
For any \(\epsilon > 0\), choose \(N = \left\lceil \frac{1}{\epsilon} - 1 \right\rceil\).
• Formal definition of convergence: \(\forall \epsilon > 0, \exists N: n > N \Rightarrow |a_n - L| < \epsilon\)
• Algebraic simplification: Combine fractions to simplify expressions
• Ceiling function: Ensures N is a natural number satisfying the condition
Divergence to infinity: \(\lim_{n \to \infty} a_n = \infty\) means for every \(M > 0\), there exists \(N\) such that \(a_n > M\) for all \(n > N\)
Formal definition: \(\forall M > 0, \exists N \in \mathbb{N}: n > N \Rightarrow a_n > M\)
We need to show: \(\forall M > 0, \exists N \in \mathbb{N}: n > N \Rightarrow \sqrt{n} > M\)
We want: \(\sqrt{n} > M\)
Squaring both sides (valid since both sides are positive): \(n > M^2\)
We need: \(n > M^2\)
Let \(N = \lceil M^2 \rceil\), where \(\lceil x \rceil\) is the ceiling function.
For any \(M > 0\), let \(N = \lceil M^2 \rceil\).
Then for all \(n > N\):
\(n > M^2\), which implies \(\sqrt{n} > \sqrt{M^2} = M\)
Therefore, \(\sqrt{n} > M\) for all \(n > N\).
Since for every \(M > 0\), we can find \(N\) such that \(\sqrt{n} > M\) for all \(n > N\), the sequence diverges to infinity.
Divergent to infinity
The sequence \(a_n = \sqrt{n}\) diverges to infinity.
For any \(M > 0\), choose \(N = \lceil M^2 \rceil\).
• Divergence to infinity definition: \(\forall M > 0, \exists N: n > N \Rightarrow a_n > M\)
• Squaring preserves inequalities: For positive numbers, \(a > b \Leftrightarrow a^2 > b^2\)
• Ceiling function: Ensures N is a natural number satisfying the condition
Oscillating sequence: A sequence that alternates between different values without approaching a limit
The sequence alternates: \(a_1 = -1\), \(a_2 = 1\), \(a_3 = -1\), \(a_4 = 1\), ...
It takes only two values: -1 and 1.
Suppose \(\lim_{n \to \infty} (-1)^n = L\) for some real number L.
Then for \(\epsilon = \frac{1}{2}\), there exists N such that \(|(-1)^n - L| < \frac{1}{2}\) for all \(n > N\).
For any odd integer \(k > N\): \(|(-1)^k - L| = |-1 - L| < \frac{1}{2}\)
For any even integer \(m > N\): \(|(-1)^m - L| = |1 - L| < \frac{1}{2}\)
From the inequalities: \(|-1 - L| < \frac{1}{2}\) and \(|1 - L| < \frac{1}{2}\)
Using the triangle inequality: \(|(-1 - L) - (1 - L)| \leq |-1 - L| + |1 - L|\)
But \(|(-1 - L) - (1 - L)| = |-1 - L - 1 + L| = |-2| = 2\)
So: \(2 \leq |-1 - L| + |1 - L| < \frac{1}{2} + \frac{1}{2} = 1\)
This gives \(2 < 1\), which is a contradiction.
Since assuming convergence leads to a contradiction, the sequence must diverge.
The sequence \(a_n = (-1)^n\) diverges.
Diverges by oscillation
The sequence \(a_n = (-1)^n\) diverges.
This is proven by contradiction: assuming convergence leads to the impossible conclusion that 2 < 1.
• Proof by contradiction: Assume convergence and derive a contradiction
• Triangle inequality: \(|a - b| \leq |a - c| + |c - b|\) for any real numbers
• Oscillation principle: If a sequence oscillates between values, it cannot converge
Convergent sequence: A sequence that approaches a finite limit as n approaches infinity
Divergent sequence: A sequence that does not converge (limit is infinite or doesn't exist)
Bounded sequence: There exists M > 0 such that \(|a_n| \leq M\) for all n
Monotonic sequence: Either always increasing or always decreasing
Cauchy sequence: Terms get arbitrarily close to each other as n increases
- Direct limit calculation: Find \(\lim_{n \to \infty} a_n\) directly
- Squeeze theorem: Bound between two convergent sequences
- Monotone convergence theorem: Bounded monotonic sequences converge
- Ratio test: For sequences involving factorials or exponentials
- Root test: For sequences with nth powers
- Contradiction: Prove divergence by assuming convergence
• Bounded Monotonic Sequence Theorem: Every bounded monotonic sequence converges
• Squeeze Theorem: If \(a_n \leq b_n \leq c_n\) and \(\lim a_n = \lim c_n = L\), then \(\lim b_n = L\)
• Cauchy Criterion: A sequence converges iff it is Cauchy
• Uniqueness of Limits: If a sequence converges, its limit is unique
• Algebraic Properties: Limits preserve arithmetic operations
Squeeze Theorem: If \(a_n \leq b_n \leq c_n\) for all n and \(\lim a_n = \lim c_n = L\), then \(\lim b_n = L\)
For all real numbers n, we know that \(-1 \leq \cos(n) \leq 1\)
Since \(n^2 > 0\) for all \(n \geq 1\), dividing the inequality by \(n^2\) preserves the direction:
\(\frac{-1}{n^2} \leq \frac{\cos(n)}{n^2} \leq \frac{1}{n^2}\)
\(\lim_{n \to \infty} \frac{-1}{n^2} = \lim_{n \to \infty} \frac{1}{n^2} = 0\)
Both bounding sequences converge to 0.
Since \(\frac{-1}{n^2} \leq \frac{\cos(n)}{n^2} \leq \frac{1}{n^2}\) and both bounding sequences approach 0,
by the Squeeze Theorem: \(\lim_{n \to \infty} \frac{\cos(n)}{n^2} = 0\)
Since the limit exists and is finite (equal to 0), the sequence converges.
The sequence \(\frac{\cos(n)}{n^2}\) converges to 0.
Convergent sequence
The sequence \(a_n = \frac{\cos(n)}{n^2}\) converges to 0.
This is proven using the Squeeze Theorem with bounding sequences \(\pm\frac{1}{n^2}\).
• Squeeze Theorem: Bounded oscillating functions divided by growing functions tend to 0
• Bounded functions: Trigonometric functions are bounded by ±1
• Limit preservation: Positive denominators preserve inequality direction
Ratio test for sequences: If \(\lim_{n \to \infty} \left|\frac{a_{n+1}}{a_n}\right| = L\), then: if \(L < 1\), \(\lim a_n = 0\); if \(L > 1\), \(\lim a_n = \infty\); if \(L = 1\), test is inconclusive
We need to find: \(\lim_{n \to \infty} \left|\frac{a_{n+1}}{a_n}\right|\)
Where \(a_n = \frac{n!}{3^n}\) and \(a_{n+1} = \frac{(n+1)!}{3^{n+1}}\)
\(\frac{a_{n+1}}{a_n} = \frac{(n+1)!}{3^{n+1}} \cdot \frac{3^n}{n!}\)
\(= \frac{(n+1)!}{n!} \cdot \frac{3^n}{3^{n+1}}\)
\(= \frac{(n+1) \cdot n!}{n!} \cdot \frac{1}{3}\)
\(= \frac{n+1}{3}\)
\(\lim_{n \to \infty} \frac{n+1}{3} = \infty\)
Since \(\lim_{n \to \infty} \left|\frac{a_{n+1}}{a_n}\right| = \infty > 1\), the sequence diverges to infinity.
Factorial functions grow faster than exponential functions.
As n increases, \(n!\) eventually dominates \(3^n\), causing the sequence to grow without bound.
\(a_n \to \infty\) (divergent)
The sequence \(a_n = \frac{n!}{3^n}\) diverges to infinity.
The ratio test shows that \(\lim \left|\frac{a_{n+1}}{a_n}\right| = \infty > 1\).
• Ratio test for sequences: If \(\lim|\frac{a_{n+1}}{a_n}| > 1\), then \(\lim a_n = \infty\)
• Factorial properties: \((n+1)! = (n+1) \cdot n!\)
• Growth rate comparison: Factorial functions grow faster than exponential functions
Subsequence convergence: A sequence converges iff all its subsequences converge to the same limit
Cauchy sequence: For every \(\epsilon > 0\), there exists N such that \(|a_m - a_n| < \epsilon\) for all \(m,n > N\)
Limsup and liminf: \(\limsup a_n = \lim_{n \to \infty} \sup_{k \geq n} a_k\), \(\liminf a_n = \lim_{n \to \infty} \inf_{k \geq n} a_k\)
Cluster points: Values that are limits of some subsequence
Convergence in probability: Statistical convergence concepts
- Root test: \(\lim \sqrt[n]{|a_n|} = L\), converges if \(L < 1\), diverges if \(L > 1\)
- Integral test: Compare with improper integral for positive, decreasing sequences
- Comparison test: Compare with known convergent/divergent sequences
- Limit comparison test: Compare ratios of sequences
- Alternating series test: For alternating sequences
- Stolz-Cesàro theorem: Discrete L'Hôpital's rule
• Bolzano-Weierstrass Theorem: Every bounded sequence has a convergent subsequence
• Stolz-Cesàro Theorem: If \(\lim \frac{a_{n+1}-a_n}{b_{n+1}-b_n} = L\) and \(b_n\) strictly monotonic, then \(\lim \frac{a_n}{b_n} = L\)
• Dominated Convergence: Conditions under which limits and integrals commute
• Monotone Convergence Theorem: Monotonic bounded sequences converge
• Cauchy Completeness: Every Cauchy sequence converges in complete spaces
Questions & Answers
Question: What's the difference between proving convergence using the formal definition versus using convergence tests?
Answer: The formal definition provides rigorous mathematical proof:
- Formal definition: Proves convergence directly from the definition: \(\forall \epsilon > 0, \exists N: n > N \Rightarrow |a_n - L| < \epsilon\)
- Convergence tests: Provide shortcuts based on established theorems (Squeeze, Ratio, Root, etc.)
Use the formal definition when:
- You need to prove convergence rigorously
- Standard tests don't apply
- Building foundational understanding
Use convergence tests when:
- Applicable conditions are met
- Quick determination is needed
- Working with complex expressions
Tests are derived from the formal definition but provide efficient shortcuts.
Question: How do I know when a sequence diverges versus when it just hasn't converged yet?
Answer: A sequence diverges if it definitively does NOT converge:
- Diverges to infinity: Grows without bound (e.g., \(n^2\))
- Oscillating divergence: Doesn't settle to any value (e.g., \((-1)^n\))
- Unbounded growth: Becomes arbitrarily large (e.g., \(n!\))
Key indicators of divergence:
- Terms grow beyond any bound
- Terms oscillate between values without settling
- Subsequences converge to different limits
- Sequence is unbounded
A sequence that hasn't converged YET might still converge later, but a divergent sequence will never converge. The mathematical proof shows definitively whether convergence is impossible.
Question: Why does the Squeeze Theorem work? What's the intuition behind it?
Answer: The Squeeze Theorem works because of the "sandwich effect":
- If \(a_n \leq b_n \leq c_n\) for all n, then \(b_n\) is "sandwiched" between \(a_n\) and \(c_n\)
- If both bounding sequences approach the same limit L, then \(b_n\) has nowhere else to go but L
Intuitively:
- Imagine two convergent sequences acting as "walls" that close in on a third sequence
- As n increases, the walls get closer together
- The middle sequence gets compressed between the walls
- Eventually, the middle sequence is forced to converge to the same limit as the walls
Mathematically: If both bounding sequences get arbitrarily close to L, then by the sandwiching, the middle sequence must also get arbitrarily close to L.
Question: What happens when the ratio test gives exactly 1? How do I proceed then?
Answer: When \(\lim \left|\frac{a_{n+1}}{a_n}\right| = 1\), the ratio test is INCONCLUSIVE:
The test doesn't provide information about convergence or divergence. Examples:
- \(a_n = \frac{1}{n}\): \(\lim \frac{a_{n+1}}{a_n} = \lim \frac{n}{n+1} = 1\), but \(\sum a_n\) diverges
- \(a_n = \frac{1}{n^2}\): \(\lim \frac{a_{n+1}}{a_n} = \lim \frac{n^2}{(n+1)^2} = 1\), but \(\sum a_n\) converges
When the ratio test is inconclusive, try:
- Root test: \(\lim \sqrt[n]{|a_n|}\)
- Comparison test with known sequences
- Integral test (if applicable)
- Limit comparison test
- More sophisticated tests like Raabe's test
Remember: Inconclusive ≠ divergent. The sequence may still converge.
Question: How do these convergence concepts apply to numerical methods and algorithms?
Answer: Convergence is fundamental to numerical methods:
- Iterative methods: Sequences of approximations must converge to the true solution
- Error analysis: Understanding how error sequences behave
- Algorithm stability: Small perturbations shouldn't cause divergence
- Rate of convergence: Determines computational efficiency
Examples:
- Newton's method: \(x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\) must converge to root
- Numerical integration: Approximation sequences must converge to true integral
- Machine learning: Optimization sequences must converge to minimum
Without convergence guarantees, numerical methods are unreliable. Convergence analysis ensures algorithms produce accurate results.