PhysicalDType

⚠️ This is a work in progress. The API is not stable and may change without notice. DO NOT use this in production code.

Overview

PhysicalDType is an experimental NumPy parametric DType for representing physical dimensions and units using NEP-41 and NEP-43.

The goal is to create a DType that can represent both physical dimension (e.g. Length, Mass, Time) and physical units (e.g. m, km, ms).

⚠️ Physical dimension is only supported for now. Physical unit support is supposed to be implemented in the near future.

How to build this repo

Only Linux is supported for now.

uv sync --extra dev
make build

Usage

import numpy as np
from physicaldtype import PhysicalDType


mass = np.array([1, 2, 3], dtype=PhysicalDType({"M": 1}))
acc = np.array([1, 2, 3], dtype=PhysicalDType({"L": 1, "T": -2}))

force = mass * acc
print(force.dtype) 
# PhysicalDType('PhysicalDimension({'L': 1.0, 'M': 1.0, 'T': -2.0, 'I': 0.0, 'Theta': 0.0, 'N': 0.0, 'J': 0.0})')

License

This project is licensed under the MIT License. See the LICENSE file for details.

References

To create a parametric DType, I referred to the following resources:

• unytdtype

  • This library creates a parametric DType for physical units utilizing the unyt library.

• numpy-quaddtype

  • This library creates a cross-platform Quad (128-bit) float Data-Type for NumPy.

Current status

PhysicalDType is currently experimental.

Supported:

• Physical dimensions represented by SI base dimensions
• NumPy array creation with PhysicalDType
• Basic arithmetic operations such as addition and multiplication

Not yet supported:

• Physical units and unit conversion
• Many NumPy ufuncs and operations
• Platforms other than Linux