Bernoulli number¶

Define a rational-number sequence by the exponential generating function for x divided by exp(x) minus one, yielding coefficients that govern power sums, Euler-Maclaurin correction, and zeta-value identities.

Version: v3 · 2026-09-19 · History
Domain-specific #: 1369
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Number Theory and Analysis, Special Number Sequences and Generating Functions → Mathematics

Core Idea¶

Under the modern first convention, the Bernoulli numbers are the rational coefficients determined by \(\frac{x}{e^x-1}=\sum_{n=0}^{\infty}B_n\frac{x^n}{n!}\), so \(B_0=1\), \(B_1=-\tfrac12\), and \(B_{2m+1}=0\) for \(m\geq1\).^[1]^[1] Multiplying the defining formal series by the series for the exponential minus one and equating coefficients yields a triangular recurrence, so each new value is fixed by earlier values; the same coefficient family enters Bernoulli polynomials and correction terms for replacing sums by integrals.

Its autonomous residual is the exact rational coefficient sequence and its normalization-dependent identities, not any number introduced by Jacob Bernoulli, Bernoulli trials, Bernoulli polynomials as functions, Euler numbers, or a generic recursively defined sequence. The identity fails when the factorial is omitted, ordinary and exponential generating functions are mixed, both signs for the first coefficient are used in one identity, odd-index vanishing is extended to index one, convergence claims replace formal-series algebra, or generalized parameters are silently set to one.

Recognition requires an analyst to write the normalization, compute the first coefficients by formal series multiplication, distinguish ordinary from exponential generating functions, state whether the alternate first Bernoulli number is used, and verify index and factorial placement in every transferred formula. Once established, it supports expressing sums of powers, formulating Euler-Maclaurin summation, relating even-index values to the Riemann zeta function, studying denominator arithmetic, and comparing generalized Bernoulli families without turning those uses into the definition.

How would you explain it like I'm…

The Recipe Number List

Bernoulli numbers are a special list of fractions that come from one fixed math recipe. Each new number in the list can be worked out from the ones before it. They are useful for shortcuts, like quickly adding up 1 + 4 + 9 + 16 and so on, without adding every number one at a time.

Numbers From a Special Recipe

The Bernoulli numbers are a famous list of numbers, mostly fractions, written B0, B1, B2 and so on. They come from one exact recipe, and each new one can be figured out from the ones before it. With the usual choice, B0 is 1, B1 is −1/2, and every odd-numbered one after that is 0. Mathematicians use them for things like quick formulas for adding up powers, such as 1² + 2² + 3² + …. Because there are a couple of versions of the recipe, you always have to say which one you're using, especially for B1.

Series Coefficients of x/(eˣ−1)

The Bernoulli numbers B_n are the rational numbers defined as the coefficients in the expansion x/(e^x − 1) = Σ B_n x^n/n!. Under this convention B_0 = 1, B_1 = −1/2, and B_n = 0 for every odd n greater than 1. Multiplying both sides by the series for e^x − 1 and matching coefficients gives a recurrence, so each value is determined by the earlier ones. They show up in formulas for sums of powers, in the Euler-Maclaurin formula that corrects the error when a sum is replaced by an integral, and in values of the Riemann zeta function at even integers. Care is needed: some sources use B_1 = +1/2, and the n! in the formula (an exponential generating function) matters.

Under the modern convention, the Bernoulli numbers are the rational coefficients Bₙ in the exponential generating function x/(eˣ − 1) = Σₙ Bₙ xⁿ/n!, giving B₀ = 1, B₁ = −1/2, and B₂ₘ₊₁ = 0 for m ≥ 1. Multiplying the series by eˣ − 1 and equating coefficients yields a triangular recurrence, so each Bₙ is determined by its predecessors. The same coefficients appear in Bernoulli polynomials and in the correction terms of the Euler–Maclaurin formula for approximating sums by integrals. Applications include closed forms for sums of powers, relations between even-index Bernoulli numbers and values of the Riemann zeta function, and the arithmetic of their denominators. Correct use depends on normalization: the generating function is exponential, not ordinary, so the factorial must be kept; the sign of B₁ differs between conventions and must not be mixed within one identity; odd-index vanishing does not apply at index one; and the definition is a formal-series identity, not a convergence claim. The Bernoulli numbers should not be confused with Bernoulli trials, Bernoulli polynomials as functions, or Euler numbers.

Structural Signature¶

Carrier: a sequence of rational numbers \((B_n)_{n\geq0}\) interpreted as coefficients of a formal or analytic exponential generating function
Inputs or antecedent state: index convention, formal variable, exponential generating function, value of the first Bernoulli number, recurrence, Bernoulli-polynomial normalization, coefficient extraction, and domain of any analytic identity
Constitutive operation: Multiplying the defining formal series by the series for the exponential minus one and equating coefficients yields a triangular recurrence, so each new value is fixed by earlier values; the same coefficient family enters Bernoulli polynomials and correction terms for replacing sums by integrals
Invariant: one declared generating-function or equivalent recurrence convention uniquely fixes every rational coefficient and explicitly resolves the sign ambiguity at index one
Recognition test: write the normalization, compute the first coefficients by formal series multiplication, distinguish ordinary from exponential generating functions, state whether the alternate first Bernoulli number is used, and verify index and factorial placement in every transferred formula
Output or consequence: expressing sums of powers, formulating Euler-Maclaurin summation, relating even-index values to the Riemann zeta function, studying denominator arithmetic, and comparing generalized Bernoulli families
Failure boundary: the factorial is omitted, ordinary and exponential generating functions are mixed, both signs for the first coefficient are used in one identity, odd-index vanishing is extended to index one, convergence claims replace formal-series algebra, or generalized parameters are silently set to one

What It Is Not¶

It is not the whole field of number theory and analysis; many objects in that field do not satisfy its constitutive rule.
It is not its canonical example. Coefficient comparison gives \(B_0=1\), \(B_1=-\tfrac12\), \(B_2=\tfrac16\), and \(B_4=-\tfrac1{30}\), while the odd values beyond \(B_1\) vanish. That is an instance, not a definition.
It is not Bernoulli polynomial. Bernoulli polynomials \(B_n(x)\) form a polynomial sequence with their own generating function; the numbers are special values such as \(B_n(0)\) under the first convention, not the full functions.
It is not an unrestricted metaphor. Two historical sign conventions differ only at the first Bernoulli number, and some authors index, name first versus second Bernoulli numbers, or evaluate polynomials at zero or one differently; formulas involving that index must declare the choice

Scope of Application¶

Bernoulli number applies when the analyst can specify a sequence of rational numbers \((B_n)_{n\geq0}\) interpreted as coefficients of a formal or analytic exponential generating function and establish that one declared generating-function or equivalent recurrence convention uniquely fixes every rational coefficient and explicitly resolves the sign ambiguity at index one. The entry states exact mathematical definitions and identities. Numerical computation, asymptotic truncation, p-adic interpolation, generalized characters, and software conventions require their own error and normalization controls.^[2]

Recognition. write the normalization, compute the first coefficients by formal series multiplication, distinguish ordinary from exponential generating functions, state whether the alternate first Bernoulli number is used, and verify index and factorial placement in every transferred formula
Comparison. Compare legitimate instances through index, first-number sign, generating-function normalization, formal or analytic reading, recurrence form, polynomial specialization, denominator, parity, zeta identity, and generalized parameter.
Boundary. Two historical sign conventions differ only at the first Bernoulli number, and some authors index, name first versus second Bernoulli numbers, or evaluate polynomials at zero or one differently; formulas involving that index must declare the choice
Use. Preserve every assumption when using the identity for expressing sums of powers, formulating Euler-Maclaurin summation, relating even-index values to the Riemann zeta function, studying denominator arithmetic, and comparing generalized Bernoulli families.

Clarity¶

A clear claim names the carrier, governing rule, assumptions, and recognition test. This matters because Bernoulli number can refer to either first-index sign convention, while Bernoulli also names a distribution, process, trial, polynomial family, and many generalized sequences. The disciplined statement is that the object counts as Bernoulli number exactly when one declared generating-function or equivalent recurrence convention uniquely fixes every rational coefficient and explicitly resolves the sign ambiguity at index one

Identity and measurement remain separate. Values are certified by exact rational arithmetic, coefficient identities, or proven recurrence; floating-point tables and pattern matching cannot establish signs, denominators, or identity compatibility. Approximation or noisy evidence may weaken a classification without changing its definition.

Manages Complexity¶

The abstraction compresses first and second conventions, Bernoulli polynomials, generalized and higher-order numbers, recurrence and explicit formulas, zeta-value expressions, denominator theorems, and computational representations into a stable carrier, rule, invariant, and failure boundary. It makes comparison tractable while retaining the variables that control validity.

Compression can hide assumptions. A responsible use therefore declares index, first-number sign, generating-function normalization, formal or analytic reading, recurrence form, polynomial specialization, denominator, parity, zeta identity, and generalized parameter and returns to the full diagnostic whenever a convention or boundary case changes.

Abstract Reasoning¶

Type the carrier. Establish a sequence of rational numbers \((B_n)_{n\geq0}\) interpreted as coefficients of a formal or analytic exponential generating function and reject examples from a different problem.
Lock the rule. Express that one declared generating-function or equivalent recurrence convention uniquely fixes every rational coefficient and explicitly resolves the sign ambiguity at index one independently of one notation or implementation.
Derive carefully. Infer expressing sums of powers, formulating Euler-Maclaurin summation, relating even-index values to the Riemann zeta function, studying denominator arithmetic, and comparing generalized Bernoulli families only under the stated assumptions.
Stress-test. Contrast the legitimate boundary case—Two historical sign conventions differ only at the first Bernoulli number, and some authors index, name first versus second Bernoulli numbers, or evaluate polynomials at zero or one differently; formulas involving that index must declare the choice—with this counterexample: the sequence of independent Bernoulli random variables from repeated coin trials shares the name but is stochastic data rather than the rational coefficient sequence defined by the generating function.

Knowledge Transfer¶

Transfer within number theory and analysis is strong when new cases preserve the same carrier, mechanism, and diagnostic. The move from Coefficient comparison gives \(B_0=1\), \(B_1=-\tfrac12\), \(B_2=\tfrac16\), and \(B_4=-\tfrac1{30}\), while the odd values beyond \(B_1\) vanish. to Euler-Maclaurin summation uses even Bernoulli numbers as coefficients of endpoint-derivative corrections when approximating a finite sum by an integral. demonstrates that continuity.^[3]

Outside the domain, only the skeleton—encode a reusable correction sequence as coefficients of one normalized generating object, letting algebraic composition determine all members and their application identities—travels automatically. The terms exponential generating function, coefficient extraction, rational number, recurrence, Bernoulli polynomial, Faulhaber formula, Euler-Maclaurin formula, zeta function, and sign convention retain domain-specific meanings, so every role and inference must be revalidated.

Examples¶

Canonical¶

Coefficient comparison gives \(B_0=1\), \(B_1=-\tfrac12\), \(B_2=\tfrac16\), and \(B_4=-\tfrac1{30}\), while the odd values beyond \(B_1\) vanish. The values are not empirical data: they are forced by the generating identity and can be recovered successively from its triangular coefficient equations.^[2] It is canonical because the carrier, rule, invariant, and consequence are all inspectable.^[1]

Mapped back: a sequence of rational numbers \((B_n)_{n\geq0}\) interpreted as coefficients of a formal or analytic exponential generating function → Multiplying the defining formal series by the series for the exponential minus one and equating coefficients yields a triangular recurrence, so each new value is fixed by earlier values; the same coefficient family enters Bernoulli polynomials and correction terms for replacing sums by integrals → one declared generating-function or equivalent recurrence convention uniquely fixes every rational coefficient and explicitly resolves the sign ambiguity at index one → expressing sums of powers, formulating Euler-Maclaurin summation, relating even-index values to the Riemann zeta function, studying denominator arithmetic, and comparing generalized Bernoulli families

Applied / In Practice¶

Euler-Maclaurin summation uses even Bernoulli numbers as coefficients of endpoint-derivative corrections when approximating a finite sum by an integral. The differentiability, endpoint, truncation, and remainder hypotheses belong to Euler-Maclaurin, while the coefficient values come from the same normalized Bernoulli sequence.^[3] It qualifies only after the same diagnostic and failure boundary are checked.^[2]

Mapped back: declared instance → recognition test → boundary check → qualified use

Structural Tensions¶

T1: Exact identity vs. practical recognition. The constitutive condition may be exact while evidence is indirect. Diagnostic: Can the reviewer state both the condition and the warrant?
T2: Canonical form vs. variants. first and second conventions, Bernoulli polynomials, generalized and higher-order numbers, recurrence and explicit formulas, zeta-value expressions, denominator theorems, and computational representations can preserve or change the identity. Diagnostic: Which named role is invariant across the variants?
T3: Compression vs. hidden assumptions. The label is useful only while prerequisites remain visible. Diagnostic: Can each downstream inference be traced to a declared assumption?
T4: Autonomy vs. reduction. The candidate uses broader structures but claims the exact rational coefficient sequence and its normalization-dependent identities, not any number introduced by Jacob Bernoulli, Bernoulli trials, Bernoulli polynomials as functions, Euler numbers, or a generic recursively defined sequence. Diagnostic: Does that residual still support independent recognition after the parent and neighbors are subtracted?

Structural–Framed Character¶

The entry is structurally mixed but domain-framed. Its portable skeleton is encode a reusable correction sequence as coefficients of one normalized generating object, letting algebraic composition determine all members and their application identities; its identity-bearing terms are exponential generating function, coefficient extraction, rational number, recurrence, Bernoulli polynomial, Faulhaber formula, Euler-Maclaurin formula, zeta function, and sign convention. Those terms determine admissible objects, evidence, and consequences inside number theory and analysis.

Structural Core vs. Domain Accent¶

The structural core is a carrier governed by Multiplying the defining formal series by the series for the exponential minus one and equating coefficients yields a triangular recurrence, so each new value is fixed by earlier values; the same coefficient family enters Bernoulli polynomials and correction terms for replacing sums by integrals and tested by write the normalization, compute the first coefficients by formal series multiplication, distinguish ordinary from exponential generating functions, state whether the alternate first Bernoulli number is used, and verify index and factorial placement in every transferred formula. The domain accent is constitutive rather than decorative, so an analogy that preserves only the skeleton is not another instance of Bernoulli number.

This entry is a kind of Recurrence.

The defining coefficient identity yields a literal triangular dependence of each indexed value on prior values, instantiating recurrence; rationality, normalization, and analytic-number-theory identities provide the specialization. What makes it its own entry is the exact rational coefficient sequence and its normalization-dependent identities, not any number introduced by Jacob Bernoulli, Bernoulli trials, Bernoulli polynomials as functions, Euler numbers, or a generic recursively defined sequence.

Relationships to Other Abstractions¶

Current abstraction Bernoulli number Domain-specific

Parents (1) — outgoing typed relations

Bernoulli number is a kind of Recurrence Prime

The defining coefficient identity yields a literal triangular dependence of each indexed value on prior values, instantiating recurrence; rationality, normalization, and analytic-number-theory identities provide the specialization.

Hierarchy path (1) — routes to 1 parentless root

Bernoulli number → is a kind of Recurrence

Neighborhood in Abstraction Space¶

Bernoulli number sits in a crowded region of the domain-specific corpus (30^th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25

Not to Be Confused With¶

Bernoulli polynomial. A polynomial-valued sequence whose endpoint specializations yield the two first-index conventions.
Euler number. A different special-number sequence associated with secant or hyperbolic-secant expansions.
Bernoulli distribution. A two-outcome probability law unrelated to the coefficient definition.
Generalized Bernoulli number. Adds a character, order, or parameter and changes the generating function and arithmetic identities.

References¶

[1] K. Dilcher, 'Bernoulli and Euler Polynomials,' Chapter 24 of the NIST Digital Library of Mathematical Functions, especially sections 24.1–24.6, current version accessed 2026-08-30, https://dlmf.nist.gov/24. registry ↩a ↩b ↩c

[2] Tom M. Apostol, 'A Primer on Bernoulli Numbers and Polynomials,' Mathematics Magazine 81(3), 178–190 (2008), DOI 10.1080/0025570X.2008.11953532. registry ↩a ↩b ↩c

[3] Ronald L. Graham, Donald E. Knuth, and Oren Patashnik, Concrete Mathematics: A Foundation for Computer Science, 2^nd ed., Addison-Wesley, 1994, sections on sums and Bernoulli numbers, ISBN 978-0-201-55802-9. registry ↩a ↩b