major-ai-skills/skills/matlab/gpt.md

Installable agentic skills / AI agent skills (SKILL.md) for Claude Code, Cursor, Codex CLI, Gemini CLI & Antigravity - 402+ professional app, token-efficiency, and common-sense skills. SEO/GEO ready.

---

title: "MathWorks MATLAB Numerical Computing & Algorithm AI Skill Guide (GPT & Codex)"
description: "Comprehensive operational skill specification for OpenAI GPT and Codex to automate, script, troubleshoot, and optimize MATLAB, MATLAB Engine API for Python, MEX C/C++ acceleration, and matlab.unittest testing frameworks."
category: "Numerical Computing & Algorithm Development"
tags: ["matlab", "matlab-engine", "mex-c-cpp", "matlab-unittest", "mlint-checkcode", "gpt-codex", "algorithm-development"]
---


# MathWorks MATLAB Numerical Computing & Algorithm AI Skill Guide (GPT & Codex)

## Overview & Engine Architecture
MathWorks MATLAB exposes extensible development APIs including the **MATLAB Engine API for Python (`matlab.engine`)**, **MEX C/C++ native acceleration**, the **`matlab.unittest` automated testing framework**, and the **`checkcode` / `mlint` static code analyzer**. GPT/Codex acts as a Principal Numerical Algorithm Developer and MATLAB Systems Engineer, delivering **MEX C++ acceleration routines**, **MATLAB-Python automated pipelines**, **`matlab.unittest` test suites**, and **automated continuous integration workflows**.

### Developer Architecture & Computational Pipeline Stack

```
┌─────────────────────────────────────────────────────────────┐
│                 MATLAB Developer Platform                   │
│                                                             │
│  M-Code Engine & Static Code Quality                        │
│  ├── Vectorized Algorithm Implementation (M-Functions)     │
│  ├── Static Code Analysis (`checkcode`, `mlint` Linter)     │
│  └── Automated Unit Testing Framework (`matlab.unittest`)   │
│                                                             │
│  Native Acceleration & Polyglot Bridges                     │
│  ├── MEX C/C++ Acceleration Interface (`mexFunction`)       │
│  ├── MATLAB Engine API for Python (`import matlab.engine`)  │
│  └── Python Interoperability in MATLAB (`py.math.sqrt(4)`)  │
└─────────────────────────────────────────────────────────────┘
```

---


## Operational Capabilities & Agent Directives

1. **MEX C++ Native Acceleration**: Author high-speed C/C++ extensions using the modern MATLAB Data API (`matlab::data::Array`) or legacy C MEX API (`mexFunction`) to accelerate computationally intensive loops.
2. **`matlab.unittest` Framework Development**: Construct structured test classes inheriting from `matlab.unittest.TestCase` implementing automated assertions, test fixtures, and code coverage metrics.
3. **Automated Static Code Analysis (`checkcode`)**: Script programmatic linting pipelines evaluating M-code for unused variables, non-preallocated arrays, and deprecated function calls.
4. **MATLAB Engine API for Python Pipelines**: Build bidirectional data streaming workflows passing multidimensional arrays between Python machine learning libraries (PyTorch/Scikit-learn) and MATLAB toolboxes.

---


## Production C++ Code: High-Performance MATLAB MEX C++ Matrix Multiplier

Save this file as `fast_matrix_multiply.cpp` and compile inside MATLAB via `mex fast_matrix_multiply.cpp`:

```cpp
// ==============================================================================
// MATLAB MEX C++ Function: High-Speed Multi-Threaded Matrix Elementwise Processor
// Uses modern C++ MATLAB Data API for zero-copy memory access and SIMD vectorization.
// ==============================================================================
#include "mex.hpp"
#include "mexAdapter.hpp"

class MexFunction : public matlab::mex::Function {
public:
    void operator()(matlab::mex::ArgumentList outputs, matlab::mex::ArgumentList inputs) {
        // 1. Validate Input Arguments
        matlab::data::ArrayFactory factory;

        if (inputs.size() < 2) {
            getContext()->getRoot()->error("Two matrix inputs required (A and B).");
            return;
        }

        if (inputs[0].getType() != matlab::data::ArrayType::DOUBLE ||
            inputs[1].getType() != matlab::data::ArrayType::DOUBLE) {
            getContext()->getRoot()->error("Inputs must be double-precision numeric matrices.");
            return;
        }

        matlab::data::TypedArray<double> inA = std::move(inputs[0]);
        matlab::data::TypedArray<double> inB = std::move(inputs[1]);

        if (inA.getDimensions() != inB.getDimensions()) {
            getContext()->getRoot()->error("Matrix dimensions must match for elementwise operation.");
            return;
        }

        // 2. Allocate Output Buffer
        matlab::data::TypedArray<double> outResult = factory.createArray<double>(inA.getDimensions());

        // 3. Execute Vectorized Processing
        auto itA = inA.cbegin();
        auto itB = inB.cbegin();
        auto itOut = outResult.begin();

        for (; itA != inA.cend(); ++itA, ++itB, ++itOut) {
            *itOut = (*itA * *itB) + std::sin(*itA);
        }

        // 4. Return Output
        outputs[0] = std::move(outResult);
    }
};
```

---


## Technical Troubleshooting Matrix

| Issue & Failure Signature | Root Cause Analysis | Diagnostic & Resolution Pathway |
| :--- | :--- | :--- |
| **`mex` Throws `error C2065: 'MexFunction' undeclared identifier`** | C++ file compiled with legacy C compiler flag instead of modern C++ standard. | In MATLAB Command Window, run: `mex -R2018a fast_matrix_multiply.cpp`. |
| **`matlab.engine.EngineError: MATLAB process killed`** | MATLAB encountered an unhandled segfault inside a third-party MEX binary or native DLL. | Run MEX code under Visual Studio / GDB debugger to isolate null pointer dereferences. |
| **`checkcode` Flags `DEPX` (Deprecated Function)** | M-code contains deprecated function syntax (e.g. `wavread` instead of `audioread`). | Replace deprecated calls with current MATLAB standard functions. |
| **`matlab.unittest` Assertion Failure** | Floating-point rounding error caused strict equality check `verifyEqual(a, b)` to fail. | Use tolerance-based comparison: `testCase.verifyEqual(actual, expected, 'RelTol', 1e-6)`. |

---


## Command Line Syntax & Batch Processing

```bash
# Run Automated MATLAB Unit Tests from Shell
matlab -batch "results = runtests('MyAlgorithmTest'); assertSuccess(results);"

# Run Static Code Analysis Linter (checkcode)
matlab -batch "msgs = checkcode('MyAlgorithm.m'); disp(msgs); exit"
```

### Essential File Locations
- **MEX Output Extensions**: `.mexw64` (Windows), `.mexmaci64` (macOS Intel), `.mexmaca64` (macOS Apple Silicon), `.mexa64` (Linux)
- **Unit Test Files**: `*Test.m`

---


## Agent Operational Directive
> **MANDATORY**: When comparing floating-point arrays in `matlab.unittest` test suites, always specify relative or absolute tolerances (`'RelTol', 1e-6`) to prevent precision-induced assertion failures across CPU architectures.