Visualize Eigenvectors & Eigenvalues | QuantumSketch
Visualize eigenvectors as the vectors that don't change direction under a matrix โ they only stretch by the eigenvalue. Watch the grid transform around them.
Visualize eigenvectors as the special vectors that keep their direction when a matrix transforms space โ they only stretch by a factor called the eigenvalue. Every other vector gets knocked off its line; eigenvectors stay on theirs.
The core idea
Apply a matrix A to the plane. Most vectors rotate to a new direction. But a few lie on lines that A merely stretches:
v is an eigenvector; ฮป (lambda) is its eigenvalue โ the stretch factor.
The animation, beat by beat
- Show the grid with several test vectors fanned out.
- Apply the matrix (e.g.
[[2,1],[1,2]]) โ the grid deforms, as a transformation. - Watch most vectors rotate off their original lines.
- Highlight the survivors โ vectors that stayed on their span, just longer.
- Label the eigenvalues โ here, stretch factors 3 and 1.
Why this matters
| Field | Eigenvectors give you | |---|---| | ML / PCA | Principal directions of data | | Physics | Vibration modes, stable states | | Google PageRank | The dominant eigenvector |
Seeing which lines survive a transformation is what makes the algebra (det(A โ ฮปI) = 0) meaningful.
Manim building blocks
NumberPlane + ApplyMatrix for the transform, highlighted Vector objects for the eigen-directions, and Text/MathTex labels for ฮป. The eigenvectors visibly hold their line while everything else swings away.
The prompt
"Apply [[2,1],[1,2]] to the plane; show most vectors rotating off their lines, highlight the eigenvectors staying on their span, labeled with eigenvalues 3 and 1."
โ Render it at quantumsketch.app. Related: Visualize Matrix Multiplication.
Written by Shihab Shahriar Antor ยท Shahriar Labs
FAQ
Q.What is an eigenvector, intuitively?
An eigenvector is a vector whose direction doesn't change when a matrix transforms the space โ it only gets stretched or shrunk. Most vectors get knocked off their original line when you apply a matrix, but eigenvectors stay on their own line, just longer or shorter (or flipped). The factor by which an eigenvector stretches is its eigenvalue. Animating a transformation and highlighting which arrows stay on their span โ while every other arrow rotates away โ makes eigenvectors instantly intuitive, instead of being just the solutions to det(A โ ฮปI) = 0.
Q.How can I animate eigenvectors of a matrix?
Describe it to an AI animation tool: 'Apply the matrix [[2,1],[1,2]] to the plane, show most vectors rotating off their lines, and highlight the eigenvectors that stay on their span, stretching by their eigenvalues 3 and 1.' QuantumSketch renders a narrated Manim animation showing the grid deform while the special directions hold. Manim's ApplyMatrix plus highlighted Vector objects make the eigenvectors visually pop against the rotating background โ all driven by your prompt.