Why Combine MATLAB and Python
MATLAB and Python have complementary strengths: MATLAB excels at rigorous numerical computing, signal processing, and its curated domain-specific toolboxes, while Python offers a vast general-purpose ecosystem for web scraping, deployment, and rapidly evolving machine-learning libraries. Rather than forcing a full rewrite in one language or the other, MATLAB provides direct interoperability in both directions, letting a project use each language for the part of the pipeline it's best suited for.
Cricket analogy: Combining MATLAB's signal-processing strength with Python's web-scraping ecosystem to build a ball-tracking analytics pipeline is like pairing a specialist spin bowler with a specialist death-overs bowler — each contributes where they're strongest.
Calling Python from MATLAB
MATLAB lets you call Python functions and access Python objects directly using the py. prefix — for example, py.numpy.array([1 2 3]) — or by running arbitrary Python statements with pyrun, after pointing MATLAB at a specific interpreter with pyenv('Version', '/path/to/python'). Values returned from Python calls arrive in the MATLAB workspace as Python-typed objects (like py.numpy.ndarray), not native MATLAB types, so they typically need explicit conversion — e.g., double(pyResult) — before you can use them in ordinary MATLAB numeric operations.
Cricket analogy: Calling py.numpy.array(runsPerOver) from MATLAB to hand ball-by-ball data to a Python NumPy function mirrors a team analyst exporting the raw scorecard to a specialist statistician's preferred software for a specific calculation.
% Calling Python from MATLAB
pyenv('Version', '/usr/bin/python3');
data = py.numpy.array([12, 45, 78, 23, 90]);
meanVal = double(py.numpy.mean(data)); % convert back to native MATLAB double
fprintf('Mean value: %.2f\n', meanVal);Calling MATLAB from Python
The MATLAB Engine API for Python, used via matlab.engine.start_matlab(), starts (or connects to) a MATLAB session that a Python script can call functions on directly, such as eng.myFunction(arg1, arg2). This is the mirror image of calling Python from MATLAB: a Python-based application defers a specific calculation to a validated, existing MATLAB function rather than reimplementing that logic in Python, and the engine session behaves like a live, running MATLAB instance the Python process is communicating with.
Cricket analogy: Starting a MATLAB engine session from Python with matlab.engine.start_matlab() to call a bowling-speed analysis function mirrors a Python-based match-day app dialing into a dedicated MATLAB analytics server for one specific calculation.
# Calling MATLAB from Python
import matlab.engine
eng = matlab.engine.start_matlab()
speed = eng.calculateBowlingSpeed(140.0, 22.5) # velocity, release angle
print(f"Estimated ball speed: {speed:.2f} km/h")
eng.quit()Data types don't map one-to-one across the boundary: a Python NumPy array returned to MATLAB arrives as a py.numpy.ndarray object, not a native double array, and must be explicitly converted (e.g., double(somePyArray)); conversely, a MATLAB matrix passed into Python via the Engine API becomes a nested matlab.double object, not a NumPy array, so wrap it with numpy.array(matlabArray) before using NumPy operations on it.
Packaging MATLAB Code for Python Deployment
MATLAB Compiler SDK's compiler.build.pythonPackage compiles a MATLAB function into a standalone Python package that end users install with pip install and call like any ordinary Python module, without needing a full MATLAB license — they only need the free, redistributable MATLAB Runtime installed. This is the deployment path of choice when a validated MATLAB algorithm needs to ship as part of a larger Python-based product, since it protects the underlying MATLAB source while making the algorithm callable from plain Python code.
Cricket analogy: Packaging a MATLAB bowling-analysis function into a standalone Python package with MATLAB Compiler SDK mirrors licensing a proprietary ball-tracking algorithm to broadcasters who don't own the underlying hardware, just the analysis output.
MATLAB Compiler SDK's compiler.build.pythonPackage can package a MATLAB function into a standalone Python package that end users install with pip install, without requiring them to own a MATLAB license — but it does require the MATLAB Runtime to be installed, and the generated package's supported Python version must match what MATLAB Compiler SDK targets for that MATLAB release.
- MATLAB and Python each have ecosystem strengths — MATLAB for numerical/signal-processing rigor and toolboxes, Python for general-purpose scripting and ML/web libraries — and interoperability lets you use both without rewriting code.
- Call Python from MATLAB using the
py.prefix (e.g.,py.numpy.array(...)) orpyrun, after configuring the interpreter withpyenv. - Data returned from Python arrives as Python-typed objects (e.g.,
py.numpy.ndarray) and must be explicitly converted to native MATLAB types likedouble. - Call MATLAB from Python using the MATLAB Engine API (
matlab.engine.start_matlab()), which starts or connects to a MATLAB session and lets you call MATLAB functions directly from Python code. - MATLAB matrices passed into Python become
matlab.doubleobjects, not NumPy arrays, and need explicit conversion withnumpy.array(...)before using NumPy operations. - MATLAB Compiler SDK's
compiler.build.pythonPackagepackages MATLAB functions into installable Python packages, letting end users run them viapip installwith the MATLAB Runtime instead of a full MATLAB license. - Always verify Python and MATLAB version compatibility (
pyenvreports the active Python version) since each MATLAB release supports only specific Python versions.
Practice what you learned
1. How do you call a Python function directly from MATLAB code?
2. What must you do with a value returned from a Python function call in MATLAB before using it in ordinary MATLAB numeric operations?
3. Which API lets you call MATLAB functions directly from a Python script?
4. When a MATLAB matrix is passed into Python via the Engine API, what type does it become in Python, and what must you do to use NumPy operations on it?
5. What does MATLAB Compiler SDK's `compiler.build.pythonPackage` allow you to do?
Was this page helpful?
You May Also Like
MATLAB Toolboxes Overview
A tour of MATLAB's major add-on toolboxes — what they add on top of base MATLAB, when to reach for each one, and how to check what's installed and licensed.
Debugging and Profiling MATLAB Code
Practical techniques for finding correctness bugs with MATLAB's interactive debugger and finding performance bottlenecks with the Profiler.
Object-Oriented Programming in MATLAB
Learn how MATLAB's classdef syntax lets you bundle data and behavior into classes, using properties, methods, inheritance, and encapsulation to build maintainable, reusable code.
Related Reading
Related Study Notes in Programming
Browse all study notesApache Spark Study Notes
Programming · 30 topics
ProgrammingApache Flink Study Notes
Programming · 30 topics
ProgrammingHadoop Study Notes
Programming · 30 topics
ProgrammingSnowflake Study Notes
Programming · 30 topics
ProgrammingApache Airflow Study Notes
Programming · 30 topics
Programmingdbt (Data Build Tool) Study Notes
Programming · 30 topics