Skip to main content

Lightweight automatic differentiation package for higher-order differentiation.

Project description

Documentation Status

njet: Lightweight automatic differentiation

A lightweight AD package, using forward-mode automatic differentiation, in order to determine the higher-order derivatives of a given function in several variables.

Features

  • Higher-order (forward-mode) automatic differentiation in several variables.
  • Support for NumPy, SymPy and mpmath.
  • Differentiation of expressions containing nested higher-order derivatives.
  • Complex differentiation (Wirtinger calculus) possible.
  • Faa di Bruno's formula for vector-valued functions implemented.
  • Lightweight and easy to use.

Installation

Install this module with pip

pip install njet

Quickstart

An example function we want to differentiate

from njet.functions import exp
f = lambda x, y, z: exp(-0.23*x**2 - 0.33*x*y - 0.11*z**2)

Generate a class to handle the derivatives of the given function (in this example up to order 3)

from njet import derive
df = derive(f, order=3)

Evaluate the derivatives at a specific point

df(0.4, 2.1, 1.73)

{(0, 0, 0): 0.5255977986928584,
 (0, 0, 1): -0.2000425221825019,
 (1, 0, 0): -0.46094926945363685,
 (0, 1, 0): -0.06937890942745731,
 (0, 0, 2): -0.03949533176976862,
 (0, 2, 0): 0.009158016044424365,
 (1, 0, 1): 0.1754372919540542,
 (0, 1, 1): 0.026405612928090252,
 (2, 0, 0): 0.1624775219121247,
 (1, 1, 0): -0.11260197000076322,
 (2, 1, 0): 0.2827794849469999,
 (1, 1, 1): 0.04285630978229049,
 (0, 1, 2): 0.005213383793609458,
 (0, 2, 1): -0.0034855409065079135,
 (0, 3, 0): -0.0012088581178640162,
 (3, 0, 0): 0.2815805411804125,
 (2, 0, 1): -0.061838944839754675,
 (0, 0, 3): 0.10305063303187477,
 (1, 2, 0): 0.03775850015116166,
 (1, 0, 2): 0.034637405962087094}

The indices here correspond to the powers of the variables x, y, z in the multivariate Taylor expansion. They can be translated to the tensor indices of the corresponding multilinear map using a built-in routine. Example:

Obtain the gradient and the Hessian

df.grad()

{(2,): -0.2000425221825019,
 (0,): -0.46094926945363685,
 (1,): -0.06937890942745731}
df.hess()

{(2, 2): -0.03949533176976862,
 (1, 1): 0.009158016044424365,
 (0, 2): 0.1754372919540542,
 (1, 2): 0.026405612928090252,
 (0, 0): 0.1624775219121247,
 (0, 1): -0.11260197000076322}

Further reading

https://njet.readthedocs.io/en/latest/index.html

License

njet: Automatic Differentiation Library

Copyright (C) 2021, 2022, 2023 by Malte Titze

njet is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

njet is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with njet. If not, see https://www.gnu.org/licenses/.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

njet-0.5.2.tar.gz (90.0 kB view details)

Uploaded Source

Built Distribution

njet-0.5.2-py3-none-any.whl (63.0 kB view details)

Uploaded Python 3

File details

Details for the file njet-0.5.2.tar.gz.

File metadata

  • Download URL: njet-0.5.2.tar.gz
  • Upload date:
  • Size: 90.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.4.2 requests/2.22.0 setuptools/45.2.0 requests-toolbelt/0.8.0 tqdm/4.30.0 CPython/3.8.10

File hashes

Hashes for njet-0.5.2.tar.gz
Algorithm Hash digest
SHA256 57f8678e6b4ee272dc88edf416a6975c5122a10a6f77fc06d6530e8cd647ea80
MD5 763af7443c26c18d6254be3b25ffc29a
BLAKE2b-256 8266142e17db802d4d66d4ec67461d96a41c4bfdfc81721505d51acd32ea2f73

See more details on using hashes here.

File details

Details for the file njet-0.5.2-py3-none-any.whl.

File metadata

  • Download URL: njet-0.5.2-py3-none-any.whl
  • Upload date:
  • Size: 63.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.4.2 requests/2.22.0 setuptools/45.2.0 requests-toolbelt/0.8.0 tqdm/4.30.0 CPython/3.8.10

File hashes

Hashes for njet-0.5.2-py3-none-any.whl
Algorithm Hash digest
SHA256 ddd0223942113c27f2047a4cd0a61eae321ce5d8288cc1289fdf51c20ee048fb
MD5 730e1764ea027075fe3b7e76afccdcf1
BLAKE2b-256 8a3d6d90c284ca4877ff843bafb1e418404deb00ea8128fa463a4be6472702f1

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page