N-Square Identity¶

An algebraic composition identity that represents a product of two sums of n squares as another sum of n squares, with its witness type and domain stated.

Version: v1 · 2026-10-03 · History
Domain-specific #: 13452
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomain: Quadratic Forms → Mathematics
Aliases: Sum-of-n-squares identity

Core Idea¶

An $n$-square identity gives $n$ output coordinates $z$ so that, for $Q_n(x)=\sum_{i=1}^n x_i^2$, the equality $Q_n(x)Q_n(y)=Q_n(z)$ holds. The type of witness matters. Bilinear polynomial coordinate laws exist in the familiar $n=1,2,4,8$ dimensions and are uniformly substitutable. Hurwitz's theorem excludes other bilinear dimensions over fields of characteristic not $2$. Pfister's power-of-two theorem supports generic rational-function identities, including $n=16$, and all-input multiplicative closure of represented field values, without creating a total bilinear $16$-tuple multiplication.[^{ref-96a9c0f3d6c3][}ref-aabb325b8d10]

Scope of Application¶

Over integers, Conrad's two-square formula turns $5=1^2+2^2$ and $13=2^2+3^2$ into $65=(-4)^2+7^2$. In a different field-theoretic setting, Pfister's construction gives a generic $16$-square identity over a rational-function field. Its matrix proof produces rational coordinate witnesses, potentially with denominators; a chosen generic formula need not be evaluable at every concrete tuple. The full theorem separately proves multiplicative closure of sums of $2^k$ squares in any field.[^ref-aabb325b8d10]

Clarity¶

The family is not one fixed formula or one normed algebra. A bilinear polynomial identity is a stronger coordinate-operation claim than a rational function-field equality. A generic rational equality differs again from an all-concrete-input closure theorem. The distinction explains why Hurwitz's $1,2,4,8$ restriction and Pfister's rational $16$-square result can both be true.[^{ref-96a9c0f3d6c3][}ref-aabb325b8d10]

Manages Complexity¶

The notation $Q_n(x)Q_n(y)=Q_n(z)$ compresses long coordinate formulas into a witness-preserving relation. For low-dimensional bilinear cases, the witness is directly computable without division. Pfister's matrix proof packages higher-dimensional field closure into scaled orthogonality of matrices, but choosing valid matrices and handling denominator-zero cases remains real work; compression is not permission to substitute into an undefined rational formula.[^{ref-96a9c0f3d6c3][}ref-aabb325b8d10]

Abstract Reasoning¶

First fix $n$, the base field or ring, and whether the desired output law must be bilinear everywhere or may be rational generically. Use the explicit bilinear formula when its dimension and hypotheses apply. For $n=2^k$ over a field, invoke Pfister's full theorem for all-value representability; if using a displayed rational formula, check its denominators before specialization. Failure of bilinearity at $n=16$ does not imply failure of sixteen-square representability under multiplication.[^{ref-96a9c0f3d6c3][}ref-aabb325b8d10]

Knowledge Transfer¶

The representation-preserving product structure travels within algebra from integer two-square calculations to rational function fields, but the witness mechanism does not transfer unchanged. Live Euler's Four-Square Identity is a narrower $n=4$ member; quaternion/normed-algebra nodes add structure; prime Closure describes a theorem consequence without defining this identity family. The frozen Pfister's sixteen-square identity is a narrower requested candidate and remains separately held; this family draft does not count it covered. A domain-general witness-composition skeleton is a future-prime question.[^{ref-96a9c0f3d6c3][}ref-aabb325b8d10]

[^ref-96a9c0f3d6c3]: Keith Conrad, “The Hurwitz Theorem on Sums of Squares”, §§1–3, original author-written mathematical proof exposition. [^ref-aabb325b8d10]: Keith Conrad, “Pfister's Theorem on Sums of Squares”, §§1–2, original author-written mathematical proof exposition.

Neighborhood in Abstraction Space¶

N-Square Identity sits in a moderately populated region (58^th percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Field Extensions & Galois-Theoretic Structures (18 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08