---
title: "Singular Value Decomposition | Linear Algebra"
description: "Singular Value Decomposition factors a matrix into U, Σ, and V* to reveal rank, geometry, and data patterns in Linear Algebra and Differential Equations."
canonical: "https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition"
type: "key-term"
subject: "Linear Algebra and Differential Equations"
unit: "Unit 13"
---

# Singular Value Decomposition | Linear Algebra

## Definition

Singular Value Decomposition, or SVD, writes a matrix as UΣV*. In Linear Algebra and Differential Equations, it is a way to break a matrix into rotation, scaling, and rotation parts so you can study its structure.

## What It Is

Singular Value Decomposition is a matrix factorization in Linear Algebra and Differential Equations that writes a matrix A as A = UΣV*. The matrices U and V are orthogonal, and Σ is a diagonal matrix whose entries are the singular values, usually listed from largest to smallest.

That factorization tells you how the matrix acts on space. One orthogonal matrix rotates or reflects vectors, Σ stretches them along special directions, and the other orthogonal matrix rotates or reflects again. So instead of looking at a matrix as a block of numbers, you can see it as a clean sequence of geometric moves.

The singular values matter because they measure how strong each direction is. A large singular value means the matrix stretches that direction a lot, while a very small singular value means that direction contributes little. If a singular value is zero, that direction gets collapsed completely, which connects SVD to rank and to whether a matrix loses information.

A useful way to think about SVD is that it finds the best coordinate directions for the matrix, not just the ones you started with. Those directions are built from the columns of V and U, and they line up with the matrix's most meaningful action. That is why SVD is often used when the original data has messy coordinates but the underlying pattern is simpler.

Here is a compact example of how this shows up. Suppose a data matrix has one very large singular value and the rest are small. That means most of the variation lives in one main direction, so you can keep that direction and drop the tiny ones for approximation or compression. In class, this is the same idea behind reducing a complicated matrix to its strongest components.

A common mistake is to confuse singular values with eigenvalues. They are related ideas, but they are not the same thing. Eigenvalues come from square matrices and describe vectors that keep their direction, while singular values exist for any matrix shape and describe how much the matrix stretches orthogonal directions.

## Why It Matters

SVD matters in this course because it connects matrix algebra, geometry, and data analysis in one move. Once you can factor a matrix into U, Σ, and V*, you can talk about what the matrix does without guessing from its raw entries.

That makes SVD useful for understanding rank, least squares ideas, and approximation. If a matrix has many tiny singular values, you can often replace it with a simpler matrix that keeps the main pattern but throws away small noisy parts. That is the logic behind compression and low-rank approximations.

SVD also shows up when the course shifts from pure algebra to applications like computer graphics and data analysis. In graphics, it can help simplify image data or transform shapes. In data analysis, it helps separate signal from noise and can reveal the main directions in a dataset.

It also gives you a better lens for reading other topics in the course. When you later see matrix decomposition, eigenvalues, or covariance matrix ideas, SVD gives you a framework for comparing how matrices act on vectors and how much information they preserve.

## Connections

### [matrix decomposition](/linear-algebra-and-differential-equations/key-terms/matrix-decomposition)

SVD is one specific kind of matrix decomposition, so it fits inside the bigger idea of breaking a matrix into simpler pieces. Other decompositions are useful for different goals, but SVD is especially good for geometry and data approximation because it separates direction and scaling so cleanly.

### Eigenvalues

Eigenvalues and singular values both describe how a matrix transforms vectors, but they answer different questions. Eigenvalues focus on directions that stay on the same line after transformation, while singular values measure stretch along orthogonal directions and work for any matrix shape, not just square ones.

### [Principal Component Analysis (PCA)](/linear-algebra-and-differential-equations/key-terms/principal-component-analysis-pca)

PCA uses the same general idea as SVD, which is why the two often show up together in data analysis. PCA looks for the directions with the most variance in a dataset, and SVD is one of the main tools used to compute those directions from a matrix.

### [covariance matrix](/linear-algebra-and-differential-equations/key-terms/covariance-matrix)

The covariance matrix organizes how variables vary together, and SVD can help extract the dominant directions in that structure. When you see a covariance matrix in data work, SVD gives a way to break it into major axes of spread, which is exactly what makes dimension reduction possible.

## On the AP Exam

A problem set or quiz question usually asks you to compute or interpret an SVD, identify the singular values, or explain what the factorization says about the matrix. You might be given a matrix and asked what its largest singular value means, which directions are most important, or why a truncated SVD is a good approximation.

If the class uses applications, you may also see a data matrix or image matrix and need to explain what happens when only the first one or two singular values are kept. The move is not just to write down U, Σ, and V*, but to connect them to rank, approximation quality, and the geometry of the transformation. If a question compares SVD to eigenvalues or matrix diagonalization, look for the fact that SVD works more broadly and focuses on orthogonal stretching directions.

## Singular Value Decomposition vs Eigenvalues

Students mix these up because both describe how matrices act on vectors and both show up in matrix analysis. The difference is that eigenvalues come from square matrices and special eigenvectors, while SVD gives singular values for any matrix and organizes the action into orthogonal rotations and stretches.

## Key Takeaways

- Singular Value Decomposition writes a matrix as UΣV*, which splits the transformation into two orthogonal motions and one scaling step.
- The singular values in Σ are listed from largest to smallest, so they show which directions matter most in the matrix.
- SVD works for rectangular matrices, which makes it broader than eigenvalue methods that require square matrices.
- Small singular values often correspond to weak directions or noise, so dropping them gives a simpler approximation of the original matrix.
- In Linear Algebra and Differential Equations, SVD is a bridge between pure matrix theory and real uses like compression, graphics, and data analysis.

## FAQs

### What is Singular Value Decomposition in Linear Algebra and Differential Equations?

Singular Value Decomposition, or SVD, is a way to factor a matrix into UΣV*. In this course, it shows how a matrix rotates, stretches, and rotates vectors again. The diagonal entries of Σ are the singular values, and they tell you which directions are strongest.

### How is SVD different from eigenvalues?

Eigenvalues describe special vectors of square matrices that keep their direction after a transformation. SVD works for any matrix shape and describes stretch along orthogonal directions instead. If you are asked to compare them, the biggest clue is that SVD is broader and more geometric.

### Why do singular values matter?

Singular values measure how much a matrix stretches each important direction. Large values show strong components of the transformation, while tiny values often point to weak structure or noise. That is why they matter for approximation and compression.

### How is SVD used in data analysis or image compression?

You keep only the largest singular values and their matching vectors to build a lower-rank approximation of the original matrix. That reduces the amount of information while preserving the main pattern. In image work, this can shrink storage without destroying the main visual features.

## Related Study Guides

- [13.4 Computer Graphics and Data Analysis](/linear-algebra-and-differential-equations/unit-13/computer-graphics-data-analysis/study-guide/mcx2d2tfQKpf6EhU)

## About This Document

Canonical Fiveable pages are available as Markdown at the same path plus `.md`.

- [llms.txt](https://fiveable.me/llms.txt): index of Fiveable's sections and URL patterns
- [llms-full.txt](https://fiveable.me/llms-full.txt): complete subject and unit listing
- [MCP server](https://fiveable.me/mcp): call Fiveable as tools instead of fetching pages (`https://fiveable.me/api/mcp`)
- [MCP server for AP teachers](https://fiveable.me/mcp/teachers): a teacher's classes, assignments and AP-rubric grading (`https://fiveable.me/api/mcp/teacher`)

## Structured Data

```json
{"@context":"https://schema.org","@graph":[{"@type":"LearningResource","@id":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition#resource","name":"Singular Value Decomposition | Linear Algebra","url":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition","learningResourceType":"Concept explainer","educationalLevel":"AP® / High School","about":{"@id":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition#term"},"audience":{"@type":"EducationalAudience","educationalRole":"student"},"dateModified":"2026-07-03T02:23:26.893Z","isPartOf":{"@type":"Collection","name":"Linear Algebra and Differential Equations Key Terms","url":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms"},"publisher":{"@type":"Organization","name":"Fiveable","url":"https://fiveable.me"}},{"@type":"DefinedTerm","@id":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition#term","name":"Singular Value Decomposition","description":"Singular Value Decomposition, or SVD, writes a matrix as UΣV*. In Linear Algebra and Differential Equations, it is a way to break a matrix into rotation, scaling, and rotation parts so you can study its structure.","url":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms/singular-value-decomposition","inDefinedTermSet":{"@type":"DefinedTermSet","name":"Linear Algebra and Differential Equations Key Terms","url":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms"}},{"@type":"FAQPage","mainEntity":[{"@type":"Question","name":"What is Singular Value Decomposition in Linear Algebra and Differential Equations?","acceptedAnswer":{"@type":"Answer","text":"Singular Value Decomposition, or SVD, is a way to factor a matrix into UΣV*. In this course, it shows how a matrix rotates, stretches, and rotates vectors again. The diagonal entries of Σ are the singular values, and they tell you which directions are strongest."}},{"@type":"Question","name":"How is SVD different from eigenvalues?","acceptedAnswer":{"@type":"Answer","text":"Eigenvalues describe special vectors of square matrices that keep their direction after a transformation. SVD works for any matrix shape and describes stretch along orthogonal directions instead. If you are asked to compare them, the biggest clue is that SVD is broader and more geometric."}},{"@type":"Question","name":"Why do singular values matter?","acceptedAnswer":{"@type":"Answer","text":"Singular values measure how much a matrix stretches each important direction. Large values show strong components of the transformation, while tiny values often point to weak structure or noise. That is why they matter for approximation and compression."}},{"@type":"Question","name":"How is SVD used in data analysis or image compression?","acceptedAnswer":{"@type":"Answer","text":"You keep only the largest singular values and their matching vectors to build a lower-rank approximation of the original matrix. That reduces the amount of information while preserving the main pattern. In image work, this can shrink storage without destroying the main visual features."}}]},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Linear Algebra and Differential Equations","item":"https://fiveable.me/linear-algebra-and-differential-equations"},{"@type":"ListItem","position":2,"name":"Key Terms","item":"https://fiveable.me/linear-algebra-and-differential-equations/key-terms"},{"@type":"ListItem","position":3,"name":"Unit 13","item":"https://fiveable.me/linear-algebra-and-differential-equations/unit-13"},{"@type":"ListItem","position":4,"name":"Singular Value Decomposition"}]}]}
```