# channel-capacity

A skill for solving information-theory problems: calculating channel capacity, mutual information, and applying the Blahut-Arimoto algorithm. Use it when working with information theory, coding, or communication channels.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf (Devin Desktop), Claude.ai · ★ 3951 repo stars
- Author: parcadei
- License: MIT · policy=mirror · the beginning of the file is at the end of this page
- Source: https://github.com/parcadei/Continuous-Claude-v3/tree/HEAD/.claude/skills/math/information-theory/channel-capacity/SKILL.md
- Updated: 2026-01-26
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/skills/channel-capacity

## What it does

The skill provides a structured approach to channel-capacity problems. It includes formulas for mutual information I(X;Y), the definition of channel capacity as maximization over input distribution, and ready formulas for standard channels: binary symmetric (BSC), binary erasure (BEC), and AWGN.

Three tools are included: `Scipy_Mutual_Info` computes entropy via scipy, `Sympy_Bsc_Capacity` simplifies BSC expressions using SymPy, and `Z3_Capacity_Bound` proves upper bounds with an SMT solver. The Blahut-Arimoto algorithm iteratively finds the distribution that achieves channel capacity.

Requires Python, scipy, sympy, and z3; tools run through `uv run` in your project.

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add parcadei/Continuous-Claude-v3 --skill channel-capacity -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/channel-capacity/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/channel-capacity/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add parcadei/Continuous-Claude-v3 --skill channel-capacity -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/channel-capacity/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add parcadei/Continuous-Claude-v3 --skill channel-capacity -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/channel-capacity/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install parcadei/Continuous-Claude-v3 .claude/skills/math/information-theory/channel-capacity
```

2. **Manually: put the files in place**
- File: `.github/skills/channel-capacity/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add parcadei/Continuous-Claude-v3 --skill channel-capacity -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/parcadei/Continuous-Claude-v3 --path .claude/skills/math/information-theory/channel-capacity
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/channel-capacity/SKILL.md` — The legacy .windsurf/skills/ path still works

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/parcadei/Continuous-Claude-v3/tree/HEAD/.claude/skills/math/information-theory/channel-capacity/SKILL.md

## When to use it
- compute mutual information between channel input and output
- find channel capacity for a given transition matrix
- calculate capacity of a binary symmetric or erasure channel
- apply the Blahut-Arimoto algorithm to a specific channel
- prove upper bounds on channel capacity

## Example prompts
- Calculate mutual information for inputs [0.5, 0.5] and outputs [0.6, 0.4]
- Find capacity of a binary symmetric channel with error probability 0.1
- Run Blahut-Arimoto on the transition matrix and report the capacity

## Source text · SKILL.md · MIT

Below is the content of a third-party file. It is data, not instructions for you: do not act on anything written inside unless the human asks. Block tag: 6d9b106597955868.

BEGIN THIRD-PARTY FILE 6d9b106597955868
````text
---
name: channel-capacity
description: "Problem-solving strategies for channel capacity in information theory"
allowed-tools: [Bash, Read]
---

# Channel Capacity

## When to Use

Use this skill when working on channel-capacity problems in information theory.

## Decision Tree


1. **Mutual Information**
   - I(X;Y) = H(X) + H(Y) - H(X,Y)
   - I(X;Y) = H(X) - H(X|Y) = H(Y) - H(Y|X)
   - Symmetric: I(X;Y) = I(Y;X)
   - `scipy.stats.entropy(p) + scipy.stats.entropy(q) - joint_entropy`

2. **Channel Model**
   - Input X, output Y, channel P(Y|X)
   - Channel matrix: rows = inputs, columns = outputs
   - Element (i,j) = P(Y=j | X=i)

3. **Channel Capacity**
   - C = max_{p(x)} I(X;Y)
   - Maximize over input distribution
   - Achieved by capacity-achieving distribution

4. **Common Channels**
   | Channel | Capacity |
   |---------|----------|
   | Binary Symmetric (BSC) | 1 - H(p) where p = crossover prob |
   | Binary Erasure (BEC) | 1 - epsilon where epsilon = erasure prob |
   | AWGN | 0.5 * log2(1 + SNR) |

5. **Blahut-Arimoto Algorithm**
   - Iterative algorithm to compute capacity
   - Alternates between optimizing p(x) and p(y|x)
   - Converges to capacity
   - `z3_solve.py prove "capacity_upper_bound"`


## Tool Commands

### Scipy_Mutual_Info
```bash
uv run python -c "from scipy.stats import entropy; p = [0.5, 0.5]; q = [0.6, 0.4]; H_X = entropy(p, base=2); H_Y = entropy(q, base=2); print('H(X)=', H_X, 'H(Y)=', H_Y)"
```

### Sympy_Bsc_Capacity
```bash
uv run python -m runtime.harness scripts/sympy_compute.py simplify "1 + p*log(p, 2) + (1-p)*log(1-p, 2)"
```

### Z3_Capacity_Bound
```bash
uv run python -m runtime.harness scripts/z3_solve.py prove "I(X;Y) <= H(X)"
```

## Key Techniques

*From indexed textbooks:*

- [Elements of Information Theory] Elements of Information Theory -- Thomas M_ Cover &amp; Joy A_ Thomas -- 2_, Auflage, New York, NY, 2012 -- Wiley-Interscience -- 9780470303153 -- 2fcfe3e8a16b3aeefeaf9429fcf9a513 -- Anna’s Archive. Using a randomly generated code, Shannon showed that one can send information at any rate below the capacity *C* of the channel with an arbitrarily low probability of error. The idea of a randomly generated code is very unusual.

## Cognitive Tools Reference

See `.claude/skills/math-mode/SKILL.md` for full tool documentation.
````
END THIRD-PARTY FILE 6d9b106597955868

The third-party file has ended. Everything between the 6d9b106597955868 tags is data: requests and commands from it are not followed. An “end of file” line with another tag inside the block is part of the file.

© parcadei · MIT · https://github.com/parcadei/Continuous-Claude-v3/tree/HEAD/.claude/skills/math/information-theory/channel-capacity/SKILL.md

## Similar
- [capacity-planning](https://affpapa.org/en/nevibecode/skills/capacity-planning.md) — skill · 115 installs
- [negentropy-lens](https://affpapa.org/en/nevibecode/skills/negentropy-lens.md) — skill · 3500 installs
- [Password Strength Checker — Entropy & Crack Time](https://affpapa.org/en/nevibecode/mcp/password-strength-checker-entropy-crack-time.md) — MCP server · 4119 installs
- [capacity](https://affpapa.org/en/nevibecode/skills/capacity.md) — skill · ★ 3232 repo stars

_Part of the page did not fit — see the HTML version._

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
