Eigenvue

The visual learning platform for understanding algorithms, AI architectures, and quantum computing.

Eigenvue provides step-by-step algorithm visualizations for classical CS, deep learning, and generative AI algorithms. Use it as a library in your Node.js projects, or as a CLI tool.

Installation

npm install eigenvue

Requires Node.js >= 18.

Quick Start

List Available Algorithms

import { list } from "eigenvue";

// List all 22 algorithms
const algorithms = list();
console.log(algorithms);

// Filter by category
const classical = list({ category: "classical" });
const deepLearning = list({ category: "deep-learning" });
const generativeAi = list({ category: "generative-ai" });

Generate Step-by-Step Data

import { steps } from "eigenvue";

// Generate steps with default inputs
const result = steps("binary-search");
console.log(`${result.steps.length} steps generated`);

// Generate steps with custom inputs
const custom = steps("binary-search", {
  array: [1, 3, 5, 7, 9],
  target: 5,
});

Each step includes state data, visual actions, and code highlights that you can use to build custom visualizations.

Open Interactive Visualization

import { show } from "eigenvue";

// Opens in your default browser
show("binary-search");

// Custom port and inputs
show("self-attention", {
  port: 8080,
  inputs: { tokens: ["I", "love", "AI"] },
});

CLI Usage

# List all algorithms
npx eigenvue list

# Filter by category
npx eigenvue list --category classical

# Open interactive visualization
npx eigenvue show binary-search

# Print step data as JSON
npx eigenvue steps binary-search

# Version and help
npx eigenvue --version
npx eigenvue --help

Available Algorithms

Classical (7)

Binary Search, Breadth-First Search, Bubble Sort, Depth-First Search, Dijkstra's Shortest Path, Merge Sort, QuickSort

Deep Learning (5)

Backpropagation, Convolution (2D), Feedforward Neural Network, Gradient Descent, Perceptron

Generative AI (5)

BPE Tokenization, Multi-Head Attention, Self-Attention, Token Embeddings, Transformer Block

API Reference

list(options?): AlgorithmInfo[]

List available algorithms with metadata.

• options.category — Filter by category: "classical", "deep-learning", "generative-ai", or "quantum".

steps(algorithmId, inputs?): StepSequence

Generate the step-by-step sequence for an algorithm.

• algorithmId — Algorithm identifier (e.g., "binary-search").
• inputs — Custom input parameters. Uses defaults from metadata if omitted.

show(algorithmId, options?): void

Launch an interactive visualization in the browser.

• algorithmId — Algorithm identifier.
• options.port — TCP port (default: auto-select).
• options.openBrowser — Whether to open the browser (default: true).
• options.inputs — Custom input parameters.

Links

• Web App
• GitHub
• Contributing

License

MIT