Concrete Numpy is an open-source library which simplifies the use of fully homomorphic encryption (FHE).

FHE is a powerful cryptographic tool, which allows computation to be performed directly on encrypted data without needing to decrypt it first.

With FHE, you can build services that preserve the privacy of the users. FHE is also great against data breaches as everything is done on encrypted data. Even if the server is compromised, in the end no sensitive data is leaked.

Main features.

• Ability to compile Python functions (that may use NumPy within) to their FHE equivalents, to operate on encrypted data
• Support for large collection of operators
• Partial support for floating points
• Support for table lookups on integers
• Support for integration with Client / Server architectures

Installation.

OS / HW	Available on Docker	Available on PyPI
Linux	Yes	Yes
Windows	Yes	Coming soon
Windows Subsystem for Linux	Yes	Yes
macOS (Intel)	Yes	Yes
macOS (Apple Silicon, ie M1, M2 etc)	Yes (Rosetta)	Coming soon

The preferred way to install Concrete Numpy is through PyPI:

pip install concrete-numpy

You can get the concrete-numpy docker image by pulling the latest docker image:

docker pull zamafhe/concrete-numpy:v0.9.0

You can find more detailed installation instructions in installing.md

Getting started.

import concrete.numpy as cnp

@cnp.compiler({"x": "encrypted", "y": "encrypted"})
def add(x, y):
    return x + y

inputset = [(2, 3), (0, 0), (1, 6), (7, 7), (7, 1), (3, 2), (6, 1), (1, 7), (4, 5), (5, 4)]

print(f"Compiling...")
circuit = add.compile(inputset)

examples = [(3, 4), (1, 2), (7, 7), (0, 0)]
for example in examples:
    result = circuit.encrypt_run_decrypt(*example)
    print(f"Evaluation of {' + '.join(map(str, example))} homomorphically = {result}")

if you have a function object that you cannot decorate, you can use the explicit Compiler API instead

import concrete.numpy as cnp

def add(x, y):
    return x + y

compiler = cnp.Compiler(add, {"x": "encrypted", "y": "encrypted"})
inputset = [(2, 3), (0, 0), (1, 6), (7, 7), (7, 1), (3, 2), (6, 1), (1, 7), (4, 5), (5, 4)]

print(f"Compiling...")
circuit = compiler.compile(inputset)

examples = [(3, 4), (1, 2), (7, 7), (0, 0)]
for example in examples:
    result = circuit.encrypt_run_decrypt(*example)
    print(f"Evaluation of {' + '.join(map(str, example))} homomorphically = {result}")

Documentation.

Full, comprehensive documentation is available at https://docs.zama.ai/concrete-numpy.

Target audience.

Concrete Numpy is a generic library that supports variety of use cases. Because of this flexibility, it doesn't provide primitives for specific use cases.

If you have a specific use case, or a specific field of computation, you may want to build abstractions on top of Concrete Numpy.

One such example is Concrete ML, which is built on top of Concrete Numpy to simplify Machine Learning oriented use cases.

Tutorials.

Various tutorials are proposed in the documentation to help you start writing homomorphic programs:

• How to use Concrete Numpy with Decorators
• Partial support of Floating Points
• How to perform Table Lookup

More generally, if you have built awesome projects using Concrete Numpy, feel free to let us know and we'll link to it!

Need support?

License.

This software is distributed under the BSD-3-Clause-Clear license. If you have any questions, please contact us at hello@zama.ai.