Every Norm Draws a Circle

2026-06-27 · Source: DataMListic · Field: Science & Research — Mathematics & Computational Sciences · Depth: Novice, quick

Summary

The concept of a "norm" defines various methods for measuring the "length" or "size" of a vector, each yielding a valid but distinct result. For instance, an arrow pointing three units across and four units up can have a straight-line Euclidean length of five, a city-block Manhattan distance of seven, or a max norm length of four, representing its largest component. These norms are unified by a formula: raising each component to power P, summing them, and taking the Pth root. Setting P=1 yields the Manhattan distance, P=2 gives the Euclidean distance (Pythagoras), and P approaching infinity results in the max norm. Crucially, asking which vectors have a length of exactly one reveals the norm's inherent geometry, drawing shapes like a circle for Euclidean, a diamond for Manhattan, and a square for the max norm, demonstrating that each norm defines its own unique "circle."

Key takeaway

For data scientists or machine learning engineers selecting distance metrics, understanding norms is crucial because the choice fundamentally alters how vector distances are calculated and perceived. Your selection of a norm—be it Euclidean, Manhattan, or Max—directly impacts the geometry of your data space, influencing clustering, classification, and optimization algorithms. Consider how different "circles" (shapes of unit vectors) might affect your model's performance and interpretability, especially in high-dimensional contexts.

Key insights

Norms provide distinct, valid methods for measuring vector size, each defining a unique geometric space.

Principles

• Norms generalize vector length measurement via a single P-parameterized formula.
• Different P values (1, 2, infinity) correspond to distinct distance metrics.
• A norm's "unit circle" reveals its fundamental geometric properties.

Topics

• Norms
• Vector Spaces
• Distance Metrics
• Euclidean Distance
• Manhattan Distance
• Max Norm
• Geometry

Best for: AI Student, General Interest

Related on AIssential

• The Geometry Underneath the Algebra — A small set of geometric ideas forms the structural foundation of classical machine learning and statistics.
• Vector Norms - Explained — Vector norms offer diverse ways to quantify vector magnitude, each with distinct mathematical and geometric properties.
• K-Means Clustering: A Deep Dive into Unsupervised Learning — K-means clustering groups data by iteratively minimizing within-cluster distances to centroids and maximizing between-cluster separation.
• On Generation in Metric Spaces — Generatability in metric spaces is scale-sensitive and metric-dependent, especially in infinite-dimensional settings.
• Cosine Similarity is Just Direction, Not Distance — Cosine similarity accurately gauges semantic similarity by focusing on vector direction, not magnitude, essential for high-dimensional embeddings.
• The most beautiful formula not enough people understand — High-dimensional geometry reveals counterintuitive properties, with n-dimensional sphere volumes peaking then becoming infinitesimal.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by DataMListic.