Arithmetic–Geometric Mean¶
The arithmetic–geometric mean is the common limit of coupled arithmetic-mean and geometric-mean iterations on two positive numbers.
Core Idea¶
For positive \(a_0,g_0\), repeatedly form \(a_{n+1}=(a_n+g_n)/2\) and \(g_{n+1}=\sqrt{a_ng_n}\), using the same current pair for both updates. The two sequences converge to one number \(M(a_0,g_0)\), their arithmetic–geometric mean. It is a limit, not either single-step mean. The gap contracts quadratically near the limit.[^ref-5b9c29578741]
Scope of Application¶
The positive-real construction is unambiguous and gives an ordered squeeze. DLMF relates \(M(1,\sqrt{1-k^2})\) to the complete first-kind elliptic integral through \(K(k)=\pi/[2M(1,\sqrt{1-k^2})]\) for \(0\le k<1\). Brent used AGM and elliptic transformations in fast high-precision algorithms, including one for π; that π algorithm needs an auxiliary correction sequence and is not the bare AGM.[ref-5b9c29578741][ref-70e2688d7402]
Clarity¶
With starting pair \((1,1/\sqrt2)\), the first updated pair is \(a_1=(1+1/\sqrt2)/2\approx0.8535533906\) and \(g_1=2^{-1/4}\approx0.8408964153\). Both coordinates must be updated from the old pair. The first arithmetic value is not the final AGM. Brent's π algorithm also updates \(t_1=1/4-(1-a_1)^2\approx0.2285533906\), producing \(\pi_1=(a_1+g_1)^2/(4t_1)\approx3.1405792505\) after one round. Complex-input extensions need explicit square-root branch choices and are not covered by the positive-real squeeze claim.[ref-5b9c29578741][ref-70e2688d7402]
Manages Complexity¶
The elliptic identity replaces one complete integral evaluation with a fast two-coordinate recurrence, but the identity and its parameter conversion remain essential. A rapidly convergent AGM of the wrong pair accurately solves the wrong integral. For π, omitting the correction state removes the information needed to recover the constant.[ref-5b9c29578741][ref-70e2688d7402]
Abstract Reasoning¶
The arithmetic update is no lower than the geometric update. For an ordered positive pair, the upper coordinate decreases and the lower increases toward a common limit. DLMF's gap recurrence \(c_{n+1}=c_n^2/(4a_{n+1})\) explains quadratic contraction. An output quantity is obtained only when a separate identity links it to that limit.[^ref-5b9c29578741]
Knowledge Transfer¶
The recurrence transfers literally to mathematical tasks with proved AGM reductions, such as complete elliptic integrals and high-precision constant algorithms. Calling any repeated compromise an “AGM” is metaphor. The Iteration edge is a composition/presupposes prerequisite for the state-carrying construction, not a claim that the limiting value is an iterative procedure; it does not imply the AGM equations or elliptic identity.
[^ref-5b9c29578741]: NIST DLMF §19.8(i), equations 19.8.1–19.8.4. [^ref-70e2688d7402]: Brent, original 1976 AGM-based algorithms, §5 full π algorithm.
Relationships to Other Abstractions¶
Current abstraction Arithmetic–Geometric Mean Domain-specific
Parents (1) — more general patterns this builds on
-
Arithmetic–Geometric Mean presupposes Iteration Prime
The AGM value presupposes coupled iteration.
Hierarchy path (1) — routes to 1 parentless root
- Arithmetic–Geometric Mean → Iteration
Neighborhood in Abstraction Space¶
Arithmetic–Geometric Mean sits in a sparse region of the domain-specific corpus (84th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Number-Theoretic Properties & Tests (20 abstractions)
Nearest neighbors
- Least Common Multiple — 0.83
- AM–GM Inequality — 0.82
- Average Order of an Arithmetic Function — 0.82
- QR Decomposition — 0.81
- Hensel's Lemma — 0.81
Computed from structural-signature embeddings · 2026-10-08