Two algebraic proofs of the transcendence of $\mathrm{e}$ based on formal power series

This paper presents two algebraic proofs of the transcendence of $\mathrm{e}$ using formal power series—one adapting the Beukers, Bézivin, and Robba approach and the other utilizing improper integrals of formal power series—while demonstrating how these methods improve upon Hilbert's classical analytical proof.

The Gist

The Big Picture: Proving $e$ is "Special"

Imagine you have a number called $e$ (Euler's number, roughly 2.718). Mathematicians have known for a long time that $e$ is transcendental.

What does that mean? It means $e$ is a "loner." You cannot make $e$ by mixing together simple whole numbers (integers) using basic math operations (addition, multiplication, and powers).

• Analogy: Think of whole numbers as Lego bricks. You can build a tower of bricks (like $2+3=5$) or a complex castle ($2^3 \times 4 = 32$). But no matter how you stack your Lego bricks, you can never build the shape of$e$. It's a unique material that doesn't exist in the Lego box.

For over a century, the standard way to prove this (Hilbert's proof) relied on calculus and integrals.

• The Problem: Calculus often deals with "uncountable sets"—basically, infinite, continuous flows of numbers (like every single point on a line). It's like trying to prove something about a river by measuring every single drop of water. It works, but it feels messy and relies on the concept of "infinity" in a very heavy way.

Klazar's Goal: He wants to prove $e$ is a "loner" without looking at the infinite river. He wants to use Formal Power Series.

• The Analogy: Instead of measuring the river, imagine you are looking at a recipe book. You don't care about the water flowing; you only care about the list of ingredients (the coefficients) and the rules for mixing them. You treat the math like a game of algebraic chess, not a physics experiment.

---

Part 1: The Old Way (Hilbert's Proof)

Hilbert's proof is like a magic trick that uses a "trap."

1. The Setup: Assume $e$ isn't a loner. Assume you can build it with Lego bricks (integers).
2. The Trap: Hilbert creates a special "net" (an integral) that catches the number $e$.
3. The Catch: When you pull the net tight, you find two things:
   • Thing A: The net is supposed to catch a whole number (an integer).
   • Thing B: But the net is also so small and tight that the number it catches must be smaller than a grain of sand (approaching zero).
4. The Contradiction: You can't have a whole number that is smaller than a grain of sand (unless it's zero, but the math proves it's not zero).
5. Result: The assumption was wrong. $e$ is transcendental.

The Flaw: To build this "net," Hilbert had to use the concept of continuous curves and infinite areas (uncountable sets). Klazar asks: Can we do this trick without the continuous river?

---

Part 2: The First New Proof (The "Recipe Book" Approach)

This proof is based on work by Beukers, Bézivin, and Robba. It treats numbers like infinite lists of ingredients.

• The Concept: Instead of a function $f(x)$ that changes as $x$ moves, we look at a Formal Power Series. Think of this as a long string of beads: $a_0 + a_1x + a_2x^2 + \dots$

  • In the old proof, $x$ was a variable moving along a line.
  • In this proof, $x$ is just a placeholder. It's a label on the bead. We only care about the numbers (the beads) themselves.

• The Strategy:

  1. Klazar assumes $e$ is built from integers.
  2. He builds a "Recipe Book" (a rational function) that describes how these integers interact.
  3. He proves that if $e$ were built from integers, this Recipe Book would have to be "perfectly rational" (simple and predictable).
  4. However, by analyzing the "beads" (the coefficients), he shows the Recipe Book is actually broken. It has a "glitch" (a pole) that shouldn't be there.
  5. The Result: Since the Recipe Book breaks, the assumption that $e$ is built from integers must be false.

• Why it's better: It never looks at a continuous line. It just counts beads and checks if the list makes sense. It's pure logic, no "infinity river" required.

---

Part 3: The Second New Proof (The "Ghost Integral")

This is Klazar's own contribution. He takes Hilbert's "trap" (the integral) and translates it into the language of the "Recipe Book."

• The Analogy: Imagine you have a machine that calculates the "weight" of a polynomial. In the old days, you had to weigh it on a scale that measured continuous fluid.

• The Innovation: Klazar invents a "Ghost Integral."

  • He defines a rule for "integrating" (summing up) these bead-lists without ever needing a continuous line.
  • He uses a "Shift" operation. Imagine sliding your list of beads to the left or right.
  • He proves a "Ghost Euler Identity": If you slide the beads of $e^{-x}$ around, they magically turn into factorials ($1, 2, 6, 24...$).

• The Execution:

  1. He sets up the same "trap" as Hilbert: $0 = \text{Part A} + \text{Part B}$.
  2. Part A (The Small Stuff): He uses his "Ghost Integral" rules to show this part is tiny (exponentially small).
  3. Part B (The Big Stuff): He uses the "Ghost Euler Identity" to show this part is a huge whole number (a multiple of $r!$).
  4. The Contradiction: Just like Hilbert, he finds a whole number that is too small to exist.

• Why it's better: It keeps the elegant structure of Hilbert's proof but removes the heavy machinery of calculus. It's like taking a complex mechanical clock and realizing it can be built entirely out of gears and springs, without needing electricity.

---

Summary: Why Should We Care?

You might ask, "Why bother? Hilbert's proof works fine."

Klazar argues that there is a philosophical beauty in purity.

• The Uncountable Set: The idea of an infinite, continuous line of numbers is a very heavy, abstract concept.
• The Countable Set: The idea of a list of integers (1, 2, 3...) is simple and concrete.

Klazar shows that we don't need the heavy, abstract "river" to prove that $e$ is special. We can prove it using only the "beads" (integers) and simple rules. It's a cleaner, more "algebraic" way of seeing the truth.

In a nutshell:
Hilbert proved $e$ is a loner by measuring the ocean.
Klazar proved $e$ is a loner by counting the grains of sand on the beach, showing that even without the ocean, the math still doesn't add up.

Technical Summary

Based on the provided paper, here is a detailed technical summary of Martin Klazar's work on two algebraic proofs of the transcendence of $e$.

1. Problem Statement

The paper addresses the classical problem of proving the transcendence of the number $e$. While the transcendence was first proven by Hermite and later refined by Hilbert, these classical proofs rely heavily on uncountable sets (specifically, real-valued functions defined on intervals like $[0, +\infty)$ and improper Riemann integrals).

The author poses a foundational question: Is it possible to prove the transcendence of $e$ without substantially using uncountable sets? The goal is to reformulate the proof using only countable structures, specifically formal power series, thereby avoiding the analytical machinery of real analysis (limits of sequences of real numbers, uncountable domains) in favor of algebraic and combinatorial methods.

2. Methodology

The paper presents two distinct algebraic approaches that replace the analytical tools of Hilbert's proof with operations on formal power series.

A. Hilbert's Classical Proof (The Baseline)

The author first reviews Hilbert's proof to establish the contradiction mechanism:

1. Assume $P = \sum_{j=0}^n a_j e^j = 0$ for integers $a_j$ (not all zero).
2. Define a polynomial $p_r(x) = x^r((x-1)\cdots(x-n))^{r+1}$.
3. Construct an integral $I_r = \int_0^\infty p_r(x)e^{-x} dx$.
4. Split the product $P \cdot I_r$ into two parts: $A_r$ (integral from $0$to$j$) and$B_r$(integral from$j$to$\infty$).
5. Show that $A_r \to 0$ rapidly (exponentially bounded) while $B_r$ is a non-zero integer multiple of $r!$.
6. The contradiction arises because $A_r + B_r = 0$ cannot hold for large $r$.
   Critique: This method uses the uncountable set of real numbers and improper integrals.

B. First Algebraic Proof: Specialization of Beukers, Bézivin, and Robba

This proof adapts the algebraic proof of the Lindemann–Weierstrass theorem to the specific case of $e$.

• Core Concept: Uses rational formal power series ($C[[x]]$).
• Key Definitions:
  • A series is rational if it can be written as $p(x)/q(x)$.
  • Rational series have a unique partial fraction decomposition involving poles.
• Mechanism:
  1. Define sequences $u_n = \sum b_j \alpha_j^n$ and $v_n = n! \sum_{r=0}^n u_r/r!$.
  2. Assume $\sum b_j e^{\alpha_j} = 0$. This implies the generating function $V(x) = \sum v_n x^n$ satisfies a specific differential equation: $(1-x)V(x) - x^2 V'(x) = \sum \frac{b_j}{1-\alpha_j x}$.
  3. Proposition 2.1: Shows that the operator $p(x)A(x) + cx^m A'(x)$ preserves or increases pole orders.
  4. Theorem 2.2: Proves that if $v_n$ grows at a specific rate ($O(A^n)$), the series $V(x)$ must be rational.
  5. Contradiction: The derived equation for $V(x)$ implies a pole structure that contradicts the properties of rational series established in Proposition 2.1.

C. Second Algebraic Proof: Semiformal Integration

This is the author's original contribution, translating Hilbert's integral proof directly into the language of formal power series.

• Core Concept: Introduction of semiformal operations on convergent formal power series ($R\{x\}$).
• Key Definitions:
  • Shift: $f(x+b)$ is defined via Taylor expansion coefficients, valid because $f$ is convergent.
  • Newton Integral: Defined formally as the difference of the primitive evaluated at bounds: $\int_u^v f = (\int f)(v) - (\int f)(u)$.
  • Improper Integral: Defined as the limit of proper integrals along a sequence, but treated algebraically within the formal system.
  • Formal Exponential: $Exp(x) = \sum \frac{x^n}{n!}$.
• Mechanism:
  1. Define the "semiformal" exponential identity: $Exp((ax)+b/a) = e^b \cdot Exp(ax)$.
  2. Translate Hilbert's integral identity $\int_0^\infty x^n e^{-x} dx = n!$ into a formal identity (Proposition 3.16) using induction and formal derivatives.
  3. Replicate the $A_r + B_r = 0$ argument using the Newton integral of $p_r(x) \cdot Exp(-x)$.
  4. Use Proposition 3.10 (Shift of integration interval) to transform the integral $\int_i^\infty p_r(x)Exp(-x) dx$ into a form where the Euler identity applies.
  5. Establish that $B_r$ is a non-zero integer multiple of $r!$ and $A_r$ is exponentially small, leading to the same contradiction as Hilbert's proof but entirely within the domain of formal power series.

3. Key Contributions

1. Elimination of Uncountable Sets: The primary contribution is demonstrating that the transcendence of $e$ can be proven without invoking uncountable sets (real numbers, continuous functions on intervals). The proofs rely solely on countable structures (integers, rational numbers, and formal power series).
2. Algebraic Translation of Analysis: The author successfully translates the analytic concepts of improper integrals, exponential decay, and integration by parts into algebraic operations (formal differentiation, shifts, and Newton integrals) on formal power series.
3. Two Distinct Proofs:
   • The first connects the problem to the broader Lindemann–Weierstrass theorem via rational series.
   • The second provides a direct, self-contained algebraic analog of Hilbert's specific proof for $e$.
4. Formal Exponential Identities: The paper rigorously establishes identities like $Exp(x+y) = Exp(x)Exp(y)$ and integration rules within the formal power series framework, ensuring they hold without analytic convergence arguments (relying instead on the fact that the series are convergent in the standard sense, allowing coefficient-wise manipulation).

4. Results

• Theorem: The number $e$ is transcendental.
• Generalization: The first proof specializes the Beukers-Bézivin-Robba proof of the Lindemann–Weierstrass theorem to the case of integer exponents.
• Contradiction: Both proofs arrive at the contradiction $A_r + B_r = 0$ where $|A_r| < 1$ and $B_r$ is a non-zero integer for sufficiently large $r$.

5. Significance

• Foundational Logic: The paper addresses a niche but significant area in mathematical logic and foundations: the distinction between proofs using countable vs. uncountable sets. It challenges the necessity of "heavy" analysis for results that seem intuitively algebraic.
• Pedagogical Value: It offers a new perspective on the transcendence of $e$, showing that the "magic" of the integral $\int x^n e^{-x} dx$ can be understood through the combinatorial properties of factorials and formal series coefficients.
• Accessibility: By removing the uncountable sets, the proofs potentially become more accessible to those working in algebraic combinatorics or logic who may not be comfortable with the full machinery of real analysis, while maintaining rigorous mathematical standards.
• Historical Context: The author notes that while the Beukers-Bézivin-Robba proof (1990) achieved this, it is under-cited. This paper aims to revive interest in these algebraic methods and provides a new, author-developed proof that specifically targets the "improper integral" aspect of Hilbert's proof.