---
id: dscribe
version: "2.1.2"
license: Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for… (full text in the JSON record)
license_treatment: permissive
maintenance: aging
---

# dscribe — A Python package for creating feature transformations in applications of machine learning to materials science.

License: permissive · Maintenance: aging · Downloads: 76.7K/mo

## What it is and what it does

DScribe is a materials-science-focused Python library that converts atomic structures into fixed-size numerical descriptors suitable for machine learning pipelines. It implements multiple descriptor types—including SOAP (Smooth Overlap of Atomic Positions), Coulomb matrix, ACSF (Atom-centered Symmetry Functions), and MBTR (Many-body Tensor Representation)—each capturing different aspects of atomic geometry and chemistry. The library wraps numpy, scipy, and scikit-learn for numerical computation and can parallelize descriptor creation across multiple processes via joblib.

Typical use involves loading or constructing atomic structures (via ASE), instantiating a descriptor class with domain-specific parameters, and calling create() to generate fixed-size feature vectors or sparse arrays. The package supports both spectrum calculation and derivative computation with respect to atomic positions, making it suitable for force-field fitting and sensitivity analysis. It targets materials scientists and machine-learning practitioners working on atomistic property prediction, structure similarity, and visualization tasks.

Use it for:

- Generate SOAP descriptors for training machine-learning models to predict material properties from atomic structure.
- Compute Coulomb matrices for similarity-based clustering or retrieval of structurally related molecules and crystals.
- Calculate descriptor derivatives to fit interatomic potentials or analyze sensitivity of predictions to atomic positions.
- Parallelize descriptor creation across multiple CPU cores when processing large datasets of structures.
- Visualize and compare atomic structures in a fixed-dimensional descriptor space for exploratory analysis.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

DScribe transforms atomic structures into fixed-size numerical fingerprints (descriptors) for use in machine learning, visualization, and similarity analysis of materials.

Yes, with conditions. DScribe is production-stable and permissively licensed, with pre-built wheels for modern Python versions and no known vulnerabilities. Install friction is moderate due to compiled dependencies and a dependency chain of 6 packages. The 321-day release gap and aging maintenance status suggest it is stable but not actively developed; suitable for established workflows but verify compatibility with your specific scipy/scikit-learn versions before adoption in new projects.

## Install

pip install dscribe
uv add dscribe
poetry add dscribe

## Installing dscribe

Before you install: Medium install friction due to compiled C/C++ components and multiple dependencies (numpy, scipy, ase, scikit-learn, joblib, sparse). Pre-built wheels available for Python 3.9–3.13 on macOS, Linux, and Windows. Last release was 321 days ago; maintenance status is aging.

License in practice: Apache License 2.0 is permissive and allows commercial use, modification, and distribution with minimal restrictions. You must include a copy of the license and state significant changes.

Quickstart:

pip install dscribe

from ase.build import molecule
from dscribe.descriptors import SOAP

water = molecule("H2O")
soap_desc = SOAP(species=["H", "O"], r_cut=5, n_max=8, l_max=6)
soap = soap_desc.create(water, centers=[0])

Requires ASE (Atomic Simulation Environment) to construct or load atomic structures; compiled C/C++ extensions require a compatible build toolchain on source installs.

Verify before relying:

- Whether the package is actively maintained or in maintenance-only mode given the 321-day gap since last release.
- Performance characteristics and scalability limits for large numbers of structures or high-dimensional descriptor spaces.
- Compatibility with recent versions of scipy and scikit-learn beyond what the fact sheet confirms.

## Package facts

- License: Apache License Version 2.0, January 2004 http://www.apache.org/licenses/ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION 1. Definitions. "License" shall mean the terms and conditions for… (full text in the JSON record) (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: aging
- Downloads: 76.7K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

atomic structure fingerprints, materials science descriptors, SOAP descriptor, coulomb matrix, machine learning materials, atomistic feature engineering, molecular descriptor generation, materials-science, descriptor-generation, machine-learning-features

[View on SkillFed](https://skillfed.io/packages/dscribe) · [View on PyPI](https://pypi.org/project/dscribe/)