Riemann zeta function Lecture 003: Analytic properties of the zeta function

Use the learning inspector for key ideas and moments, or open the reading tabs for the complete notes.

Chapters

Learning script

Generated from the video's visuals and explanation; not verbatim speech.

0:00The clip opens with a title card naming the topic as analytic properties of the zeta function in the half-plane $\operatorname{Re}(s)>1$.

0:03The first content slide defines the object of study directly: for complex $s$ with $\operatorname{Re}(s)>1$, $\zeta(s)=\sum_{n=1}^{\infty}\frac{1}{n^s}$. The speaker reads this formula and states that the episode will examine analytic properties of $\zeta(s)$ in this region.

0:20The lecture then announces the first major step of the argument. Instead of trying to work all the way up to the boundary line $\operatorname{Re}(s)=1$, it fixes a positive constant $a$ and targets uniform convergence on the closed half-plane $\operatorname{Re}(s)\ge 1+a$. The spoken explanation ties this choice to avoiding the blow-up near real part one.

0:54Next, the slide introduces the complex-analysis tool that will convert convergence into regularity: if a sequence of analytic functions $f_n$ converges uniformly to $f$, then $f$ is analytic in a domain. Alongside that theorem, the lecture gives an informal picture of analyticity by listing consequences such as continuity and infinite differentiability.

1:23A new slide titled Convergence defines uniform convergence in words as convergence of a sequence of functions to a limit function at the same rate across the entire region. This is still an informal definition rather than a formal epsilon-based statement.

1:37The abstract notion is then specialized to the zeta series. The sequence under consideration is the partial sums $f_n(s)=\sum_{k=1}^{n}\frac{1}{k^s}$. The speaker explicitly explains the notational shift: $n$ now labels the sequence element, while $k$ is the dummy summation index inside the finite sum.

2:09To make the definition concrete, the slide writes out the first four members of the sequence: $f_1(s)=1$, $f_2(s)=1+\frac{1}{2^s}$, $f_3(s)=1+\frac{1}{2^s}+\frac{1}{3^s}$, and $f_4(s)=1+\frac{1}{2^s}+\frac{1}{3^s}+\frac{1}{4^s}$. These are exactly the truncations of the zeta series after one, two, three, and four terms.

2:34The final seconds switch to a comparison slide with two headings, pointwise convergence of $f_1,f_2,\ldots$ to $f$ and uniform convergence of $f_1,f_2,\ldots$ to $f$. The speaker says they will describe what uniform convergence is, but the actual side-by-side explanation has not yet begun within this excerpt.

2:40The segment opens on a slide titled “Convergence,” with two parallel headings: “Pointwise convergence of $f_1$, $f_2$, ..., to f” on the left and “Uniform convergence of $f_1$, $f_2$, ..., to f” on the right. The lecturer says the easiest way to understand the new topic is to compare it with pointwise convergence, which students may already have encountered informally.

2:58The left column is then built in order. First comes “First pick a point x.” Next comes “Then pick $ε > 0$.” Finally the slide asks: “Is there an index N such that whenever $n \ge N$, |$f(x) - f_n(x)$| < ε?” The spoken explanation matches this sequence: fix a particular input point, choose a tolerance, and then look for a threshold index that makes the function values close to the limit at that fixed point.

3:43Attention shifts to the right column for uniform convergence. The lecturer explicitly notes that the order changes. The slide now begins with “First pick $ε > 0$,” and then asks whether there is an index N such that whenever $n \ge N$, the inequality |$f(x) - f_n(x)$| < ε holds “for all x in a region?” The same algebraic inequality appears, but the quantifier structure is different because x has not been fixed in advance.

4:13The phrase “for all x in a region?” is visually emphasized, underscoring the universal quantifier over the whole domain. The lecturer calls this the key difference: in the uniform case, one single N must work simultaneously for every x in the region, rather than separately at each point.

4:23Two summary bullets appear at the bottom to crystallize the contrast. On the left: “N can be chosen dependent on x.” On the right: “N must be chosen for all x simultaneously.” This is the operational distinction between the two definitions: pointwise convergence allows the threshold to vary with the point, while uniform convergence forbids that dependence.

4:37The final bullets apply the definitions to the Riemann zeta function on the real domain $x>1$. The left note says, “We already showed this is true for $ζ(x)$ when $x > 1$,” referring back to an earlier integral-test argument. The right note says, “As we'll see, this is false for $ζ(x)$ when $x > 1$.” Thus the clip ends by setting up a concrete example where pointwise convergence holds but uniform convergence on the whole region $x>1$ does not.

5:20We begin by defining the partial sums of the Riemann zeta function: $f_1(x) = 1$, $f_2(x) = 1 + 1/2^x$, and in general, $f_n(x) = \sum_{k=1}^n 1/k^x$. As established previously, this sequence of functions converges pointwise to $\zeta(x)$ for all $x > 1$.

5:41Next, we investigate whether this convergence is uniform on the interval (1, $\infty$). To disprove uniform convergence, we must find an $\epsilon > 0$ such that for any N, there exists an $n > N$ and an $x > 1$ where |$\zeta(x) - f_n(x)$| $\ge \epsilon$. We choose $\epsilon = 0.005$.

6:03First, observe that for any $x > 1$, each term $1/k^x$ in the partial sum $f_n(x)$ is less than or equal to 1. Since there are n terms, we have the upper bound $f_n(x) \le n$.

6:31The key insight is to exploit the behavior of $\zeta(x)$ near its singularity at $x=1$. We choose a specific x that depends on n: $x = 1 + 1/(n+1)$. As n grows, this x gets arbitrarily close to 1 from the right.

6:46Using a result from the integral test, we know that $\zeta(1 + 1/(n+1)) > n+1$. This provides a lower bound for the limit function at our chosen x.

7:01Now we evaluate the difference: |$\zeta(x) - f_n(x)$|. Substituting our bounds, we get |$\zeta(x) - f_n(x)$| > ($n+1$) - $n = 1$. Since $1 > 0.005 = \epsilon$, the condition for uniform convergence fails.

7:44Therefore, the sequence $f_n(x)$ does not converge uniformly to $\zeta(x)$ on the entire domain $x > 1$. The root cause is that we were able to pick an x arbitrarily close to 1, where the gap between the partial sum and the limit function becomes large.

7:51However, if we restrict our attention to a smaller domain, specifically $x \ge 1+a$ for some fixed $a > 0$, we stay away from the singularity. On this restricted interval, the sequence $f_n(x)$ does converge uniformly to $\zeta(x)$.

8:00The segment begins by analyzing the uniform convergence of the partial sums of the Riemann zeta function, $f_n(x)$, for real values $x > 1$. A graph illustrates the function $y = \zeta(x)$ and a partial sum $y = 1 + \frac{1}{2^x}$. As x approaches 1, $\zeta(x)$ exhibits a vertical asymptote, growing arbitrarily large. The instructor explains that this singularity prevents uniform convergence on the entire domain $x > 1$. A specific counterexample is provided: by choosing $\epsilon = 0.005$ and selecting $x = 1 + \frac{1}{n+1}$, the value of $\zeta(x)$ can be made greater than $n + 1$, while the partial sum $f_n(x)$ is bounded by n. This results in |$\zeta(x) - f_n(x)$| > $1 > 0.005$, violating the condition for uniform convergence.

8:41To address this, the lecture transitions to complex analysis and formally defines uniform convergence for a sequence of complex functions $f_n(s)$ converging to $f(s)$ on a region A. The definition states that for any given $\epsilon > 0$, there must exist an integer N such that for all $n \ge N$, the inequality |$f(z) - f_n(z)$| < $\epsilon$ holds simultaneously for all points z in A. The instructor then defines the partial sums $f_n(s) = \sum_{k=1}^n \frac{1}{k^s}$ and introduces the concept of tail sums, or remainder sums, denoted as $R_n(s)$. The tail sum is defined as the difference between the full zeta function and the partial sum: This formula needs review. Please report it using the page feedback control.infty $\frac{1}{k^s}$. The goal is now to prove that $f_n(s)$ converges uniformly to $\zeta(s)$ on a restricted region $\text{Re}(s) \ge 1 + a$ for some fixed $a > 0$, which effectively keeps the domain away from the singularity at $s = 1$. This is achieved by showing that for any $\epsilon > 0$, there exists an N such that |$R_n(s)$| < $\epsilon$ for all $n \ge N$ and all s in the region.

10:11The final part of the segment focuses on algebraically manipulating the tail sum $R_n(s)$ to facilitate the proof. The complex variable s is expressed in its standard form, $s = \sigma$ + it, where $\sigma$ is the real part and t is the imaginary part. Substituting this into the tail sum yields This formula needs review. Please report it using the page feedback control.infty $\frac{1}{k^{\sigma+it}}$. Using the properties of exponents, the denominator is split into $k^\sigma k$^{it}, resulting in This formula needs review. Please report it using the page feedback control.infty $\frac{1}{k^\sigma} k$^{-it}. The term k is then rewritten as $e^{\log k}$, transforming the expression to This formula needs review. Please report it using the page feedback control.infty $\frac{1}{k^\sigma} (e^{\log k})$^{-it}. Finally, applying the power of a power rule for exponents, the expression simplifies to This formula needs review. Please report it using the page feedback control.infty $\frac{1}{k^\sigma} e$^{-it($\log k$)}. This manipulation separates the magnitude and phase components of the terms in the tail sum, setting the stage for bounding the sum to prove uniform convergence.

10:40The clip opens on a slide titled “Zeta in the Complex Plane.” The first mathematical move is purely notational: the complex argument is written as $s=\sigma+it$. With that substitution, the tail of the Dirichlet series is displayed as $R_n(s)=\sum_{k=n+1}^{\infty}1/k^s$. The speaker stresses that this is not the full sum from $k=1$, but a tail beginning at $k=n+1$.

10:48The next lines rewrite the summand step by step. First $k^s$ becomes $k^{\sigma+it}$, then it splits as $k^{\sigma}k^{it}$. Moving the oscillatory factor to the numerator gives $k^{-it}$. Because the speaker says “log base e” is being used, $k$ is rewritten as $e^{\log k}$, so $(e^{\log k})^{-it}=e^{-it(\log k)}$. Thus the tail takes the form $\sum_{k=n+1}^{\infty}k^{-\sigma}e^{-it(\log k)}$, separating size from phase.

10:56A new slide assumes $\sigma>1$ and estimates the modulus of this complex tail. The key inequality is the triangle inequality: the absolute value of a sum is at most the sum of the absolute values. Applying it termwise gives $|R_n(\sigma+it)|\le \sum |k^{-\sigma}e^{-it(\log k)}|$. Since $k^{-\sigma}$ is positive real and $|e^{-it(\log k)}|=1$, the right-hand side collapses to $\sum k^{-\sigma}=R_n(\sigma)$. The speaker’s verbal image is that the terms are vectors spiraling around rather than all pointing in the same direction, so cancellation can only make the true modulus smaller.

11:39The argument then shifts from complex inputs to real inputs. Under the hypothesis “For real $\sigma\ge 1+a$,” the slide writes out the first few terms of $R_n(1+a)$ and $R_n(\sigma)$ and compares them term by term. Because $\sigma\ge 1+a$ and the bases $k$ are positive, one has $k^{\sigma}\ge k^{1+a}$. Taking reciprocals reverses the inequality, yielding $1/k^{\sigma}\le 1/k^{1+a}$. Summing these inequalities over the same index set gives $R_n(\sigma)\le R_n(1+a)$.

12:30The final slide combines the two previous estimates and adds a geometric picture. On the left, the complex plane is drawn with Real and Imaginary axes, the vertical boundary line at $1+a$, and a shaded half-plane consisting of all points with $\operatorname{Re}(s)\ge 1+a$. A sample point $\sigma+it$ is marked inside this region, with its projection $\sigma$ on the real axis. On the right, the inequalities are chained: $|R_n(\sigma+it)|\le R_n(\sigma)$ and $R_n(\sigma)\le R_n(1+a)$. By transitivity, every point in the half-plane satisfies $|R_n(s)|\le R_n(1+a)$. Since equality is attained at the boundary point $s=(1+a)+0i$, the slide concludes that the greatest value of $|R_n(s)|$ in the region $\operatorname{Re}(s)\ge 1+a$ occurs there.

13:20The lecture resumes with a slide titled 'Zeta in the Complex Plane'. The speaker notes the similarity to a previous episode, clarifying that the current focus is on the remainder term $R_n$ rather than the full zeta function $\zeta$.

13:32The goal is to complete the proof regarding the convergence of the series. The speaker recalls that the partial sums $f_n(s)$ are already known to converge pointwise to $\zeta(s)$ for $Re(s) > 1$, a result established via the integral test in a prior session.

13:44To bridge pointwise and uniform convergence, the speaker fixes a specific point on the real axis: $s = 1 + a$, where $a > 0$. By the definition of pointwise convergence, for any arbitrary $\epsilon > 0$, there exists an integer $N$ such that for all $n \ge N$, the difference $|\zeta(1+a) - f_n(1+a)|$—which equals the remainder $|R_n(1+a)|$—is strictly less than $\epsilon$.

14:12Crucially, the speaker proposes using this same $\epsilon$ and the same threshold $N$ for the entire region defined by $Re(s) \ge 1 + a$. The logic is that if the error is small at the boundary 'worst-case' point, it should be small everywhere inside the region.

14:20The derivation proceeds by taking an arbitrary complex number $s = \sigma + it$ in the region, meaning $\sigma \ge 1 + a$. The magnitude of the remainder $|R_n(\sigma + it)|$ is bounded. First, it is noted that $|R_n(\sigma + it)| \le R_n(\sigma)$, implying the error is maximized when the imaginary part is zero. Second, since $\sigma \ge 1 + a$ and the remainder function decreases as the real argument increases, $R_n(\sigma) \le R_n(1 + a)$.

14:51Combining these inequalities yields $|R_n(\sigma + it)| \le R_n(\sigma) \le R_n(1 + a) < \epsilon$. Since $|R_n(\sigma + it)|$ represents the distance between the partial sum and the limit function $\zeta(s)$, this inequality holds for all $s$ in the region $Re(s) \ge 1 + a$ using the single $N$ derived earlier.

14:59This satisfies the formal definition of uniform convergence. The speaker concludes that the sequence of partial sums $f_n(s)$ converges uniformly to $\zeta(s)$ on the region $Re(s) \ge 1 + a$.

15:14The slide transitions to a new topic: 'Complex Analysis'. The speaker introduces the Weierstrass M-test as a standard theorem that offers a faster route to proving uniform convergence, though he chose the manual derivation to build intuition.

15:25The Weierstrass M-test is stated: If you have an infinite series of functions $\sum f_n(z)$ defined on a region $A$, and if each function's magnitude is bounded by a constant $M_n$ (i.e., $|f_n(z)| \le M_n$ for all $z \in A$), and if the series of these constants $\sum M_n$ converges, then the original series of functions converges uniformly on $A$.

15:52The speaker reflects that this test could have been applied directly to the zeta function problem to skip the detailed bounding steps, but emphasizes the value of understanding the underlying mechanics of convergence definitions.

16:00The lecturer begins by referencing the Weierstrass M-test, a theorem from complex analysis used to establish uniform convergence of infinite series of functions. The slide displays the formal statement: if the absolute value of each function $|f_n(z)|$ is bounded by a constant $M_n$, and the series of these constants $\sum M_n$ converges, then the original series $\sum f_n(z)$ converges uniformly in the specified region.

16:07Addressing the question of why uniform convergence is so important, the speaker explains that it grants permission to perform critical calculus operations. Specifically, it allows mathematicians to safely interchange the order of limits with summations, integrations, and differentiation, which is essential for manipulating complex series.

16:24The focus shifts to the concept of analytic functions. The speaker defines 'analytic' as a condition significantly stronger than mere continuity or differentiability. In the context of complex analysis, an analytic function is exceptionally well-behaved, implying it is infinitely differentiable and can be locally represented by a convergent power series.

16:40To illustrate, several common analytic functions are listed on the slide. These include polynomials like $z^5 - z^2 + 1$, the trigonometric function $\sin(z)$, and the exponential function $e^z$. The lecturer also points out that the individual terms of the Riemann zeta series, $n^{-s}$, are analytic functions of the complex variable $s$.

17:16The pivotal tool for the upcoming proof is introduced: the Uniform Limit Theorem. The slide states that if a sequence of analytic functions $f_n$ converges uniformly to a limit function $f$ within a domain, then that limit function $f$ must also be analytic in that domain.

17:32Applying this theorem to the Riemann zeta function, the lecturer constructs a logical proof. First, the partial sums $f_n(s) = \sum_{k=1}^n \frac{1}{k^s}$ are identified as finite sums of analytic functions, making them analytic themselves. Second, because the series converges uniformly on regions where $Re(s) \ge 1 + a$ (for any $a > 0$), the Uniform Limit Theorem dictates that the limit function, $\zeta(s)$, is analytic on these regions.

18:07Since the parameter $a$ can be chosen to be arbitrarily small, the union of all such regions covers the entire open half-plane where $Re(s) > 1$. Consequently, the proof concludes that the Riemann zeta function is analytic everywhere to the right of the vertical line $x=1$ in the complex plane.

18:26The segment ends with a visual summary. A diagram of the complex plane is shown, with a green shaded region extending to the right of the dashed vertical line at $Re(s)=1$. This visually reinforces the main takeaway: the zeta function is highly regular and analytic throughout this specific domain.

18:40The right side of the screen directly presents the key conclusion: The function zeta of s is analytic, that is, very well behaved, on the region where the real part of s is greater than one. Here, s is a complex variable, and the real part of s denotes its real component, so the condition real part of s is greater than one describes the open right half-plane in the complex plane lying to the right of the vertical line real part of s equals one.

18:44The left-side diagram geometrizes this condition: the horizontal axis is labeled Real, the vertical axis is labeled Imaginary, the origin is zero, the point one is marked on the horizontal axis, and a vertical dashed line passes through one. The green shaded region is exactly where the real part of s is greater than one. A small red dot moves within this region to represent the complex point s currently under consideration.

18:48The narrator then outlines the course plan: in the next few videos, we will first discuss zeta's behavior in this region—that is, study zeta of s within the known analytic right half-plane. The purpose of this step is to first understand the function in a region where it has favorable properties, before tackling the more challenging extension problem.

18:52Next, the narrator notes that several more videos will be devoted to this topic before zeta is truly extended to the rest of the complex plane. Correspondingly, in the animation, the red dot crosses the dashed line into the left non-green region, visually representing the shift from the region where the real part of s is greater than one to the rest of the complex plane.

18:55Thus, although this segment is brief, it clearly establishes two layers: first, zeta of s is analytic where the real part of s is greater than one; second, the upcoming study will first focus on this region, then address extension to the full complex plane. The video does not prove analyticity here, but treats it as the starting point for the next series of discussions.

Knowledge cards