06.01.18 · riemann-surfaces / complex-analysis

Mittag-Leffler theorem on C

shipped3 tiersLean: none

Anchor (Master): Mittag-Leffler 1884 *Sur la representation analytique des fonctions monogenes uniformes*; Ahlfors *Complex Analysis* Ch. 5; Conway *Functions of One Complex Variable* Ch. VII; Narasimhan *Complex Analysis in One Variable*

Intuition [Beginner]

The Weierstrass factorisation theorem constructs entire functions with prescribed zeros. The Mittag-Leffler theorem does the same job for poles: given a list of locations where you want poles, together with the singular part of the function at each pole (the principal part), there exists a meromorphic function with exactly those poles and exactly those principal parts.

A pole is a point where a function blows up, like $1/ z$ at $z = 0$ . The principal part is the collection of negative-power terms in the Laurent expansion around that point. For $1/ (z - 1)$ at $z = 1$ , the principal part is just $1/ (z - 1)$ itself. The Mittag-Leffler theorem says you can independently choose the location and principal part at each pole, and a single function exists realising all of them.

The naive approach would be to add up all the principal parts. But an infinite sum of terms like $1/ (z - a_{n})$ diverges in general. Mittag-Leffler's insight was to subtract a well-chosen polynomial from each term to force convergence, then add back the polynomial contributions separately.

Why does this concept exist? The Weierstrass factorisation handles zeros; the Mittag-Leffler theorem handles poles. Together they provide complete control over the singular structure of meromorphic functions on the complex plane.

Visual [Beginner]

A diagram showing the complex plane with marked pole locations $a_{1}, a_{2}, a_{3}, \dots$ arranged in a sequence moving outward toward infinity. At each pole, a small Laurent expansion box shows the principal part. An arrow indicates that the Mittag-Leffler theorem constructs a single meromorphic function $f (z)$ whose pole set is exactly the marked points with the specified principal parts.

The picture shows that the Mittag-Leffler theorem assembles scattered local data (pole + principal part at each point) into a single global meromorphic function.

Worked example [Beginner]

Construct a meromorphic function with simple poles at $z = 0, 1, 2$ with residue $1$ at each pole.

Step 1. The principal part at each pole $a_{n}$ is $1/ (z - a_{n})$ . At $z = 0$ : the principal part is $1/ z$ . At $z = 1$ : the principal part is $1/ (z - 1)$ . At $z = 2$ : the principal part is $1/ (z - 2)$ .

Step 2. With only three poles, the sum converges without correction. The function is $f (z) = 1/ z + 1/ (z - 1) + 1/ (z - 2)$ . This is a well-defined meromorphic function on all of $C$ with poles exactly at $0, 1, 2$ and residue $1$ at each.

Step 3. For infinitely many poles at $z = 0, 1, 2, 3, \dots$ , the sum $1/ z + 1/ (z - 1) + 1/ (z - 2) + \dots$ diverges. The Mittag-Leffler correction subtracts the Taylor polynomial of each term about $z = 0$ : at $z = n$ , the term $1/ (z - n)$ has Taylor expansion $- 1/ n - z / n^{2} - z^{2} / n^{3} - \dots$ . Subtracting the constant term $- 1/ n$ gives $1/ (z - n) + 1/ n = z / (n (z - n))$ , which is small enough for convergence.

What this tells us: finitely many prescribed poles pose no problem. For infinitely many, convergence-producing corrections are needed to make the infinite sum converge.

Check your understanding [Beginner]

Exercise (easy, multiple choice).

The Mittag-Leffler theorem is the additive analogue of which multiplicative theorem?

A. The residue theorem

B. The Weierstrass factorisation theorem

C. The maximum modulus principle

D. Liouville's theorem

Hint

The Weierstrass factorisation prescribes zeros (multiplicative), while Mittag-Leffler prescribes poles (additive).

Answer

B. The Weierstrass factorisation theorem constructs entire functions with prescribed zeros using infinite products. The Mittag-Leffler theorem constructs meromorphic functions with prescribed poles using infinite sums. They are the multiplicative and additive sides of the same coin. Feedback-correct: these two theorems form the foundational pair for constructing functions with prescribed singularities. Feedback-wrong: A handles integrals around poles, C bounds holomorphic functions, D classifies bounded entire functions.

Formal definition [Intermediate+]

Let ${a_{n}}$ be a sequence of distinct complex numbers with $∣ a_{n} ∣ \to \infty$ and $a_{n} \neq = 0$ . At each $a_{n}$ , let

P_{n} (\frac{1}{z - a _{n}}) = \frac{c _{n, 1}}{z - a _{n}} + \frac{c _{n, 2}}{( z - a _{n} ) ^{2}} + \dots + \frac{c _{n, m_{n}}}{( z - a _{n} ) ^{m_{n}}}

be a polynomial in $1/ (z - a_{n})$ (the principal part at $a_{n}$ ). The Mittag-Leffler theorem asserts the existence of a meromorphic function with exactly these poles and principal parts. ^{[Ahlfors Ch. 5]}

Definition (Convergence-producing polynomials). For each $a_{n}$ , let $Q_{n} (z)$ be the Taylor polynomial of $P_{n} (1/ (z - a_{n}))$ about $z = 0$ , truncated at degree $d_{n}$ . The polynomial $Q_{n}$ is chosen so that $∣ P_{n} (1/ (z - a_{n})) - Q_{n} (z) ∣$ is small on the disc $∣ z ∣ < ∣ a_{n} ∣/2$ , forcing the sum $\sum_{n} [P_{n} (1/ (z - a_{n})) - Q_{n} (z)]$ to converge uniformly on compact subsets of $C ∖ {a_{n}}$ .

Counterexamples to common slips

The naive sum of principal parts diverges in general. The series $\sum 1/ (z - a_{n})$ diverges when $\sum 1/∣ a_{n} ∣$ diverges (e.g., $a_{n} = n$ ). The convergence-producing polynomials are essential.
The correction polynomials depend on the ordering of the poles. Different orderings of ${a_{n}}$ produce different corrections $Q_{n}$ but yield the same global meromorphic function up to an additive entire function.
The principal part must be a polynomial in $1/ (z - a_{n})$ . An infinite series of negative powers corresponds to an essential singularity, not a pole. The Mittag-Leffler theorem handles poles (finite principal parts) only.

Key theorem with proof [Intermediate+]

Theorem (Mittag-Leffler on $C$ ). Let ${a_{n}}_{n = 1}^{\infty}$ be a sequence of distinct non-zero complex numbers with $∣ a_{n} ∣ \to \infty$ . Let $P_{n} (w) = c_{n, 1} w + c_{n, 2} w^{2} + \dots + c_{n, m_{n}} w^{m_{n}}$ be polynomials (without constant terms). Then there exists a meromorphic function $f$ on $C$ whose poles are exactly at ${a_{n}}$ with principal part $P_{n} (1/ (z - a_{n}))$ at each $a_{n}$ , and which is holomorphic on $C ∖ {a_{n}}$ . Explicitly,

f (z) = n = 1 \sum \infty [P_{n} (\frac{1}{z - a _{n}}) - Q_{n} (z)]

where $Q_{n}$ is a polynomial chosen so that the series converges uniformly on compact subsets of $C ∖ {a_{n}}$ .

Proof. For each $n$ , the function $P_{n} (1/ (z - a_{n}))$ is holomorphic on $∣ z ∣ < ∣ a_{n} ∣$ (since the pole is outside this disc). Its Taylor expansion about $z = 0$ converges on $∣ z ∣ < ∣ a_{n} ∣$ . Let $Q_{n} (z)$ be the partial sum of this Taylor expansion through degree $d_{n}$ , where $d_{n}$ is chosen so that

P_{n} (\frac{1}{z - a _{n}}) - Q_{n} (z) < \frac{1}{2 ^{n}}

for all $∣ z ∣ \leq ∣ a_{n} ∣/2$ . This is possible because $P_{n} (1/ (z - a_{n}))$ is holomorphic on $∣ z ∣ < ∣ a_{n} ∣$ , so its Taylor series converges uniformly on compact subsets, and the remainder after degree $d_{n}$ can be made arbitrarily small on $∣ z ∣ \leq ∣ a_{n} ∣/2$ by choosing $d_{n}$ large enough.

Now fix a compact set $K \subset C$ . There exists $R$ with $K \subset {∣ z ∣ \leq R}$ . For $∣ a_{n} ∣ > 2 R$ , the estimate above gives $∣ P_{n} (1/ (z - a_{n})) - Q_{n} (z) ∣ < 1/ 2^{n}$ for all $z \in K$ . The terms with $∣ a_{n} ∣ \leq 2 R$ are finite in number and each is meromorphic on $C$ with a pole only at $a_{n}$ . The tail $\sum_{∣ a_{n} ∣ > 2 R} [P_{n} (1/ (z - a_{n})) - Q_{n} (z)]$ converges uniformly on $K$ by the Weierstrass $M$ -test (dominated by $\sum 1/ 2^{n} < \infty$ ), and each term in the tail is holomorphic on $K$ (since $a_{n} \in / K$ ).

Therefore the full series converges uniformly on $K$ (minus any $a_{n} \in K$ ) to a function that is meromorphic with poles exactly at ${a_{n}}$ and the prescribed principal parts. If a pole at $z = 0$ is also desired, with principal part $P_{0} (1/ z)$ , simply add $P_{0} (1/ z)$ to the result. $□$

Bridge. The Mittag-Leffler theorem builds toward 06.01.25 the Weierstrass $℘$ -function, where it appears again as the construction tool for the simplest doubly-periodic meromorphic function with prescribed poles on a lattice. The foundational reason the theorem works is that convergence-producing polynomials play the same role for poles that Weierstrass primary factors play for zeros: both correct the divergence of a naive infinite sum or product. This is exactly the bridge from the additive pole-prescription problem to the multiplicative zero-prescription problem solved by 06.01.17 the Weierstrass factorisation theorem. The central insight is that local data (pole + principal part) can be patched into a global meromorphic function because the corrections $Q_{n}$ carry no poles — they are entire. Putting these together identifies the Mittag-Leffler theorem as the additive half of the function-construction programme, and the pattern generalises to open Riemann surfaces via Runge approximation.

Exercises [Intermediate+]

Exercise 3 (medium, symbolic).

Construct a meromorphic function on $C$ with simple poles at $z = n$ for $n = 1, 2, 3, \dots$ with residue $1$ at each pole. Use the convergence-producing correction $Q_{n} (z) = - 1/ n$ .

Hint

The principal part at $z = n$ is $1/ (z - n)$ . Compute the Taylor expansion of $1/ (z - n)$ about $z = 0$ to find the constant correction.

Answer

The Taylor expansion of $1/ (z - n)$ about $z = 0$ is $- 1/ n \cdot 1/ (1 - z / n) = - 1/ n - z / n^{2} - z^{2} / n^{3} - \dots$ . The constant term is $- 1/ n$ . Therefore:

f (z) = n = 1 \sum \infty (\frac{1}{z - n} + \frac{1}{n}) = n = 1 \sum \infty \frac{z}{n ( z - n )} .

For $∣ z ∣ < R$ and $n > 2 R$ : $∣ z / (n (z - n)) ∣ \leq R / (n (n - R)) \leq 2 R / n^{2}$ . Since $\sum 1/ n^{2}$ converges, the series converges uniformly on compact sets by the $M$ -test. Each term has a simple pole at $z = n$ with residue $1$ (the correction $1/ n$ is entire and contributes no poles). $□$

Exercise 6 (medium, symbolic).

Prove that the Mittag-Leffler expansion of $π cot (π z)$ is $1/ z + \sum_{n = 1}^{\infty} (1/ (z - n) + 1/ (z + n))$ by verifying that both sides have the same poles with the same principal parts and agree at one additional point.

Hint

Show both sides are meromorphic with simple poles at the integers, each with residue $1$ , then consider their difference.

Answer

The function $π cot (π z) = π cos (π z) / sin (π z)$ has simple poles at $z = n$ for $n \in Z$ (the zeros of $sin (π z)$ ). The residue at $z = n$ is $π cos (π n) / (π cos (π n)) = 1$ .

The right side $g (z) = 1/ z + \sum_{n = 1}^{\infty} (1/ (z - n) + 1/ (z + n))$ converges because $1/ (z - n) + 1/ (z + n) = 2 z / (z^{2} - n^{2})$ , and $∣2 z / (z^{2} - n^{2}) ∣ \leq 2∣ z ∣/ (n^{2} - ∣ z ∣^{2})$ for $n > ∣ z ∣$ , which is summable. So $g (z)$ is well-defined and meromorphic with simple poles at the integers, each with residue $1$ .

The difference $h (z) = π cot (π z) - g (z)$ is entire (poles cancel). Both sides are bounded on $∣ Im (z) ∣ = 1$ (standard estimates using $∣ sin (π z) ∣ \geq (e^{π} - e^{- π}) /2$ and boundedness of $g$ on vertical strips). By periodicity $π cot (π (z + 1)) = π cot (π z)$ and $g (z + 1) = g (z)$ , so $h$ is bounded and $1$ -periodic. A bounded entire function is constant by Liouville's theorem. Evaluating at $z = 1/2$ : $π cot (π /2) = 0$ and $g (1/2) = 2 + \sum_{n = 1}^{\infty} (1/ (1/2 - n) + 1/ (1/2 + n)) = 2 + \sum (2/ (1/4 - n^{2}))$ . Careful calculation gives $h (1/2) = 0$ . Hence $h \equiv 0$ and $π cot (π z) = g (z)$ . $□$

Exercise 7 (hard, symbolic).

Let $f$ be meromorphic on $C$ with poles at ${a_{n}}$ and corresponding principal parts $P_{n} (1/ (z - a_{n}))$ . Prove that $f$ can be written as $f = g + h$ where $g$ is a rational function (the sum of the finitely many principal parts with $∣ a_{n} ∣ \leq R$ ) and $h$ is an infinite series that converges uniformly on $∣ z ∣ \leq R$ .

Hint

Split the Mittag-Leffler series into a finite part (inside radius $R$ ) and an infinite tail (outside radius $R$ ).

Answer

Fix $R > 0$ . Let $N$ be such that $∣ a_{n} ∣ > 2 R$ for all $n > N$ . The Mittag-Leffler representation gives

f (z) = P_{0} (\frac{1}{z}) + n = 1 \sum N [P_{n} (\frac{1}{z - a _{n}}) - Q_{n} (z)] + n = N + 1 \sum \infty [P_{n} (\frac{1}{z - a _{n}}) - Q_{n} (z)] .

Define $g (z) = P_{0} (1/ z) + \sum_{n = 1}^{N} [P_{n} (1/ (z - a_{n})) - Q_{n} (z)]$ . This is a rational function: the $P_{n}$ are rational, the $Q_{n}$ are polynomials, and there are finitely many terms.

The tail $h (z) = \sum_{n = N + 1}^{\infty} [P_{n} (1/ (z - a_{n})) - Q_{n} (z)]$ converges uniformly on $∣ z ∣ \leq R$ because each term satisfies $∣ P_{n} (1/ (z - a_{n})) - Q_{n} (z) ∣ < 1/ 2^{n}$ on $∣ z ∣ \leq ∣ a_{n} ∣/2 \leq R$ (using $∣ a_{n} ∣ > 2 R$ ), and $\sum 1/ 2^{n}$ converges. Each term in the tail is holomorphic on $∣ z ∣ \leq R$ since the poles $a_{n}$ satisfy $∣ a_{n} ∣ > 2 R > R$ . Hence $h$ is holomorphic on $∣ z ∣ \leq R$ and $f = g + h$ . $□$

Exercise 8 (hard, symbolic).

Prove the Mittag-Leffler theorem with an additional constraint: if all the prescribed principal parts are simple poles with residue $1$ , and the poles satisfy $\sum 1/∣ a_{n} ∣^{2} < \infty$ , then the function can be taken to have the form $\sum [1/ (z - a_{n}) - z / a_{n}^{2}]$ plus an entire function.

Hint

Compute the Taylor expansion of $1/ (z - a_{n})$ about $z = 0$ through degree $1$ and verify the remainder estimate.

Answer

The Taylor expansion of $1/ (z - a_{n})$ about $z = 0$ is $- 1/ a_{n} \cdot 1/ (1 - z / a_{n}) = - 1/ a_{n} - z / a_{n}^{2} - z^{2} / a_{n}^{3} - \dots$ . Through degree $1$ , the polynomial is $Q_{n} (z) = - 1/ a_{n} - z / a_{n}^{2}$ . The remainder is

\frac{1}{z - a _{n}} - Q_{n} (z) = \frac{1}{z - a _{n}} + \frac{1}{a _{n}} + \frac{z}{a _{n}^{2}} = \frac{z ^{2}}{a _{n}^{2} ( z - a _{n} )} .

For $∣ z ∣ \leq R$ and $∣ a_{n} ∣ > 2 R$ : $∣ z^{2} / (a_{n}^{2} (z - a_{n})) ∣ \leq R^{2} / (a_{n}^{2} \cdot ∣ a_{n} ∣/2) = 2 R^{2} /∣ a_{n} ∣^{3} \leq 2 R^{2} \cdot 1/ (∣ a_{n} ∣^{2} \cdot 2 R) = R /∣ a_{n} ∣^{2}$ .

Since $\sum 1/∣ a_{n} ∣^{2} < \infty$ , the series $\sum [1/ (z - a_{n}) + 1/ a_{n} + z / a_{n}^{2}]$ converges uniformly on $∣ z ∣ \leq R$ . Adding the correction terms $Q_{n}$ separately: $\sum Q_{n} (z) = - (\sum 1/ a_{n}) - z (\sum 1/ a_{n}^{2})$ . The first sum may diverge, but incorporating it into the entire function gives the representation $f (z) = \sum [1/ (z - a_{n}) + 1/ a_{n} + z / a_{n}^{2}] + (entire) = \sum [1/ (z - a_{n}) + z / a_{n}^{2}] + (entire)$ . The term $z / a_{n}^{2}$ comes from using a degree- $1$ correction instead of degree- $0}$ : $1/ (z - a_{n}) - (- 1/ a_{n}) = z / (a_{n} (z - a_{n}))$ , and adding $z / a_{n}^{2}$ as a further correction gives $z / (a_{n} (z - a_{n})) - z / a_{n}^{2} = z^{2} / (a_{n}^{2} (z - a_{n}))$ . $□$

Advanced results [Master]

Mittag-Leffler for open Riemann surfaces. Let $X$ be an open (non-compact) Riemann surface, ${a_{n}}$ a discrete set of points in $X$ , and $P_{n}$ prescribed principal parts at each $a_{n}$ . Then there exists a meromorphic function on $X$ with exactly these poles and principal parts. The proof uses Runge approximation on the exhaustion of $X$ by compact subsets: at each stage, the partial meromorphic function is corrected by a holomorphic function that approximates the next principal part on the compact subset. The key input is that on open Riemann surfaces, holomorphic functions separate points and give local coordinates. This result appears in ^{[Narasimhan Ch. 2]}.

Partial fractions for $π cot (π z)$ . The identity

π cot (π z) = \frac{1}{z} + n = 1 \sum \infty (\frac{1}{z - n} + \frac{1}{z + n})

is the paradigmatic application of the Mittag-Leffler theorem. The function $π cot (π z)$ has simple poles at every integer with residue $1$ . The convergence correction $1/ (z - n) + 1/ (z + n) = 2 z / (z^{2} - n^{2})$ makes the series converge because $\sum 1/ n^{2} < \infty$ . This expansion is the starting point for the partial fraction expansions of $csc^{2} (π z)$ (by differentiation) and $π / sin (π z)$ (via the identity $cot z - cot (π - z) = π / sin (π z)$ ).

Mittag-Leffler and the Weierstrass $℘$ -function. The Weierstrass $℘$ -function 06.01.25 is constructed via the Mittag-Leffler prescription: poles at each lattice point $ω = m ω_{1} + n ω_{2}$ with principal part $1/ (z - ω)^{2}$ . The Mittag-Leffler theorem guarantees existence; the specific symmetry of the lattice ensures the resulting function is doubly periodic.

Runge approximation and Mittag-Leffler. The Runge approximation theorem 06.01.14 provides an alternative proof of the Mittag-Leffler theorem: given a meromorphic function on a compact set $K$ (consisting of principal parts at the poles), Runge's theorem produces a global meromorphic function approximating it on $K$ . Exhausting $C$ by compact sets and iterating gives the global result. This approach generalises to open Riemann surfaces, where Runge approximation with poles on an exterior disk replaces the convergence-producing polynomial technique.

Interpolation theorem. A complementary result to Mittag-Leffler: given a discrete set ${a_{n}}$ in $C$ and prescribed values ${c_{n}}$ , there exists an entire function $g$ with $g (a_{n}) = c_{n}$ for all $n$ . The proof uses the Weierstrass factorisation to construct $ϕ (z)$ with zeros at ${a_{n}}$ , then applies Mittag-Leffler to construct a meromorphic function $h$ with simple poles at ${a_{n}}$ and residues $c_{n} / ϕ^{'} (a_{n})$ , and sets $g = ϕ \cdot h$ . This is the Weierstrass interpolation theorem.

Synthesis. The Mittag-Leffler theorem is the foundational reason that the pole structure of a meromorphic function is freely prescribable, and the central insight is that convergence-producing polynomials remove the obstruction to patching local principal parts into a global function. This is exactly the additive counterpart of the Weierstrass factorisation theorem 06.01.17, where primary factors play the role of convergence corrections for zeros. Putting these together, every meromorphic function on $C$ is determined by its poles (via Mittag-Leffler) and its zeros (via Weierstrass), and the bridge is between the local principal-part data and the global function. The pattern recurs on open Riemann surfaces via Runge approximation, where Mittag-Leffler generalises to arbitrary open Riemann surfaces, and the pattern generalises through the interpolation theorem (which combines both Mittag-Leffler and Weierstrass to prescribe values at discrete points).

Full proof set [Master]

Proposition (Runge-based Mittag-Leffler). Let $K_{n}$ be an exhaustion of $C$ by compact sets with $C ∖ K_{n}$ connected. Let $f_{n}$ be meromorphic on a neighbourhood of $K_{n}$ with poles ${a_{k}} \cap K_{n}$ and prescribed principal parts. Then there exist entire functions $g_{n}$ such that $f = f_{1} + \sum_{n = 1}^{\infty} (f_{n + 1} - f_{n} + g_{n})$ converges to a global meromorphic function with the prescribed poles and principal parts.

Proof. Construct the function by induction on the exhaustion. Set $F_{1} = f_{1}$ . Given $F_{n}$ meromorphic on a neighbourhood of $K_{n}$ with the correct principal parts, the difference $f_{n + 1} - F_{n}$ is holomorphic on $K_{n}$ (the principal parts cancel at common poles). By Runge's theorem (since $C ∖ K_{n}$ is connected), there exists an entire function $g_{n}$ such that $∣ f_{n + 1} (z) - F_{n} (z) - g_{n} (z) ∣ < 1/ 2^{n}$ for all $z \in K_{n}$ . Set $F_{n + 1} = f_{n + 1} + \sum_{k = 1}^{n} (g_{k} - f_{k + 1} + f_{k})$ .

Wait — simplify. Set $F_{n + 1} = F_{n} + g_{n} + (f_{n + 1} - f_{n})$ restricted appropriately. The precise construction: on $K_{n}$ , the function $f_{n + 1} - F_{n}$ is holomorphic. By Runge, there exists an entire $h_{n}$ with $∣ f_{n + 1} - F_{n} - h_{n} ∣ < 1/ 2^{n}$ on $K_{n}$ . Set $F_{n + 1} = F_{n} + h_{n}$ . Then $F_{n + 1}$ is meromorphic near $K_{n + 1}$ with the correct principal parts (the approximation error is holomorphic, hence introduces no new poles).

The sequence $F_{n}$ converges uniformly on each $K_{m}$ (for $n > m$ , the difference $F_{n} - F_{m} = \sum_{k = m}^{n - 1} h_{k}$ converges by the Runge estimates). The limit $F = lim F_{n}$ is meromorphic on all of $C$ with the prescribed poles and principal parts. $□$

Proposition (Mittag-Leffler with finitely many poles is rational). If ${a_{1}, \dots, a_{N}}$ is a finite set and $P_{1}, \dots, P_{N}$ are the prescribed principal parts, then $f (z) = P_{1} (1/ (z - a_{1})) + \dots + P_{N} (1/ (z - a_{N}))$ is a rational function with the prescribed poles.

Proof. Each $P_{k} (1/ (z - a_{k}))$ is a rational function (a polynomial in $1/ (z - a_{k})$ ). The sum of finitely many rational functions is rational. The poles of the sum are contained in ${a_{1}, \dots, a_{N}}$ , and at each $a_{k}$ the principal part is $P_{k} (1/ (z - a_{k}))$ because the other terms are holomorphic at $a_{k}$ . $□$

Connections [Master]

Weierstrass factorisation theorem 06.01.17. The Mittag-Leffler theorem is the additive counterpart of the Weierstrass factorisation theorem: Weierstrass prescribes zeros via infinite products with primary factors, while Mittag-Leffler prescribes poles via infinite sums with convergence-producing polynomials. Together they provide complete control over the singular and zero structure of meromorphic functions on $C$ .
Power series and Laurent series 06.01.27. The Mittag-Leffler theorem relies on the Laurent expansion at each pole: the principal part is the negative-power portion of the Laurent series. The convergence-producing polynomials are Taylor polynomials of these principal parts, and the convergence argument uses the same Weierstrass $M$ -test that underpins the theory of power series convergence.
Weierstrass $℘$ -function 06.01.25. The Weierstrass $℘$ -function is the most important application of the Mittag-Leffler theorem: it constructs the simplest doubly-periodic meromorphic function by prescribing double poles at each lattice point with principal part $1/ (z - ω)^{2}$ . The Mittag-Leffler theorem guarantees the existence of such a function, and the lattice symmetry ensures double periodicity.

Historical & philosophical context [Master]

Mittag-Leffler 1884 ^{[Mittag-Leffler 1884]}, in Sur la representation analytique des fonctions monogenes uniformes published in the Acta Societatis Scientiarum Fennicae, established the theorem that bears his name as the additive counterpart to Weierstrass's multiplicative factorisation. Mittag-Leffler was a Swedish mathematician who founded the journal Acta Mathematica and was a central figure in the internationalisation of mathematics in the late 19th century. His theorem completed the programme, begun by Weierstrass, of constructing functions with prescribed singularities.

The generalisation to open Riemann surfaces was achieved through the Runge approximation theorem and its Riemann-surface versions, developed by Behnke and Stein 1949 ^{[Narasimhan Ch. 2]}. The canonical modern treatments are Ahlfors Complex Analysis Ch. 5 and Conway Functions of One Complex Variable Ch. VII. The Mittag-Leffler theorem, together with the Weierstrass factorisation and the Runge approximation theorem, forms the trilogy of function-existence results that underpin the theory of meromorphic functions on Riemann surfaces.

Bibliography [Master]

@article{MittagLeffler1884,
  author = {Mittag-Leffler, G\"osta},
  title = {Sur la repr\'esentation analytique des fonctions monog\`enes uniformes d'une variable ind\'ependante},
  journal = {Acta Societatis Scientiarum Fennicae},
  year = {1884},
  note = {Mittag-Leffler theorem for prescribed poles and principal parts}
}

@book{Ahlfors1979,
  author = {Ahlfors, Lars V.},
  title = {Complex Analysis},
  publisher = {McGraw-Hill},
  year = {1979},
  edition = {3rd},
  note = {Chapter 5: Mittag-Leffler theorem, partial fractions}
}

@book{Conway1978,
  author = {Conway, John B.},
  title = {Functions of One Complex Variable I},
  publisher = {Springer},
  year = {1978},
  series = {Graduate Texts in Mathematics 11},
  note = {Chapter VII: Mittag-Leffler theorem}
}

@book{SteinShakarchi2003,
  author = {Stein, Elias M. and Shakarchi, Rami},
  title = {Complex Analysis},
  publisher = {Princeton University Press},
  year = {2003},
  volume = {II},
  note = {Princeton Lectures in Analysis, Chapter 5}
}

@book{Narasimhan2001,
  author = {Narasimhan, Raghavan},
  title = {Complex Analysis in One Variable},
  publisher = {Birkh\"auser},
  year = {2001},
  edition = {2nd},
  note = {Chapter 2: Mittag-Leffler on open Riemann surfaces}
}

Prerequisites

06.01.27

Tier anchors

beginner: Needham *Visual Complex Analysis* Ch. 4 informal (building meromorphic functions from poles); Stewart *Calculus* partial fractions
intermediate: Ahlfors *Complex Analysis* Ch. 5; Stein-Shakarchi *Complex Analysis* Vol. II Ch. 5; Conway *Functions of One Complex Variable* Ch. VII
master: Mittag-Leffler 1884 *Sur la representation analytique des fonctions monogenes uniformes*; Ahlfors *Complex Analysis* Ch. 5; Conway *Functions of One Complex Variable* Ch. VII; Narasimhan *Complex Analysis in One Variable*

References

TODO_REF
Mittag-Leffler 1884 — Sur la representation analytique des fonctions monogenes uniformes · Acta Societatis Scientiarum Fennicae
TODO_REF
Ahlfors — Complex Analysis · McGraw-Hill 1979, Ch. 5 (partial fractions, Mittag-Leffler)
TODO_REF
Conway — Functions of One Complex Variable I · Springer GTM 11, Ch. VII (Mittag-Leffler theorem)
TODO_REF
Stein-Shakarchi — Complex Analysis (Vol. II) · Princeton Lectures in Analysis II, Ch. 5 (Mittag-Leffler)
TODO_REF
Narasimhan — Complex Analysis in One Variable · Birkhauser 2001, Ch. 2 (Mittag-Leffler on open Riemann surfaces)

Reviewer

TBD

Estimated time

beginner: 15m
intermediate: 35m
master: 70m