Skip to content

Scientific Computing System (CDS)

Computational science, from scratch.

CDS is an open-source platform for research, simulation and discovery: 34 subpackages spanning quantum simulation, FFT, linear algebra, statistics, ODE/PDE, symbolic math, machine learning, uncertainty quantification and NLP, implemented in readable pure Python with zero runtime dependencies.

Get started Take the tour API reference

pip install scientific-computing-system
cds modules          # see what's inside

Start here

Getting Started

Install, run your first simulation, and learn the CLI in about five minutes.

Quick Start Tutorial

A guided first session: constants, statistics, a quantum circuit, an ODE.

Tour of Numerical Methods

How the solvers actually work: quadrature, RK45, LU, FFT, with worked output.

Cookbook

Problem-oriented recipes: pick the task, copy the snippet.

API Reference

All 34 public subpackages, generated from the source docstrings.

Architecture

Module dependency graph and data flow, for contributors and auditors.

Why Pure Python?

The trade-off stated plainly, and when to reach for the NumPy-based 2.0 project instead.

New to scientific Python?

CDS is designed as a first stop: every algorithm is readable pure Python, so you learn how things work, not just how to call them. Follow the tutorials in order: Quick Start, Statistics, Machine Learning, then branch out.

What people use CDS for

CDS is designed for searchable, practical scientific-Python tasks such as numerical methods, ODE and PDE solving, Monte Carlo simulation, statistics and hypothesis testing, uncertainty quantification, sensitivity analysis, dimensional analysis, signal processing, linear algebra, symbolic mathematics, and reproducible computational research.

Key Features

  • Pure Python: Every module is implemented from scratch using the Python standard library. No heavy dependencies like NumPy or SciPy required.
  • Quantum Simulation: Full state-vector simulation for single and multi-qubit circuits with entanglement and O(1) sampling.
  • Advanced Mathematics: O(N³) Partial Pivoting LU decomposition, SVD, QR, Cholesky, and adaptive ODE solvers (RK45).
  • Scientific Assurance: Uncertainty propagation, local & variance-based global sensitivity, identifiability analysis, validation checks, cross-method verification, assumption-gated causal estimation, and SI dimensional analysis.
  • Reproducible Orchestration: Approval-gated workflow execution that fails closed, plus run manifests, canonical hashing, and checkpoints.
  • Hypothesis Engine: Built-in tools for generating and statistically validating scientific hypotheses, complemented by effect-size measures (Cohen's d, Cramér's V) that quantify the magnitude of an effect alongside its significance.
  • High Reliability: Comprehensive test suite with 100% blended code coverage (statement + branch) on the reference CI cell. See the CI and codecov badges in the README for live status.
  • Interactive Tools: cds CLI (pure stdlib) and an optional Streamlit dashboard.

Overview of Modules

src/cds/ contains 34 subpackages plus causal.py and sensitivity.py — 36 public feature modules. This table summarises them and the API Reference documents each one. (cds modules prints a curated 26-entry subset, not the full catalog.)

Module Description
cds.core Shared data models (Domain, Hypothesis, HypothesisStatus)
cds.hypothesis Structured hypothesis generation and evaluation
cds.knowledge Concept graphs, research notes, ranked structured retrieval
cds.stats Descriptive stats, regression, hypothesis testing, effect sizes, time series
cds.probability Probability distributions, quantiles, seeded sampling
cds.bayes Bayesian conjugate updates, credible intervals, Bayes factors
cds.math_utils Calculus + linear algebra (SVD, QR, Cholesky, LU, eigen)
cds.interpolate interp1d / interp2d over sampled domains
cds.numerical_integration Trapezoid, Simpson, Romberg, Gauss-Legendre, adaptive, 2-D
cds.montecarlo π estimation, Monte-Carlo integration, random walks
cds.diffeq Euler, RK4, adaptive RK45, implicit stiff, symplectic
cds.pde Finite-difference heat & wave equation solvers (1-D)
cds.optimization Gradient descent, Newton, Adam, Nelder-Mead, annealing, constrained
cds.ml MLP, k-NN, k-means, CART, random forest, boosting, PCA (educational)
cds.signals DFT/FFT, convolution, Butterworth design, STFT, spectral tools
cds.wavelets Haar DWT/IDWT, multi-level decomposition, denoising
cds.quantum Single & multi-qubit state-vector simulation
cds.scientific Physical constants & classical physics formulas
cds.genetics GC content, k-mers, reverse complement, Needleman-Wunsch
cds.fractals Mandelbrot, Julia, Barnsley fern, Sierpinski triangle
cds.infotheory Entropy, cross-entropy, KL / JS divergence, mutual information
cds.modeling Symbolic expressions, differentiation, LaTeX, MathModel, fitting
cds.graph BFS, DFS, Dijkstra, Kruskal MST, topological sort, cycles
cds.uncertainty Analytic + correlated Monte-Carlo uncertainty propagation
cds.sensitivity Local & variance-based global sensitivity, identifiability
cds.validation Validation checks, data adequacy, drift / OOD reports
cds.causal Assumption-gated causal effect estimators
cds.units SI quantities, conversions, dimensional compatibility
cds.workflow Approval-gated scientific workflow orchestration
cds.provenance Run manifests, canonical SHA-256, decisions, checkpoints
cds.data_analysis CSV/tabular loading, normalisation, ASCII visualisation
cds.data_io Streaming I/O, online moments, optional HDF5 / NetCDF
cds.tools Lazy discovery + adapters for NumPy/SciPy/sklearn/SymPy/Z3
cds.plot Optional matplotlib charts via the [plot] extra
cds.nlp Educational NLP from scratch (BPE, embeddings, attention, MiniGPT)
cds.cli The cds console script (pure-stdlib argparse)

Import behaviour

Every module in the table is reachable as cds.<name>:

import cds

cds.pde  # attribute access works
cds.interpolate  # so does this
from cds.pde import solve_heat  # and the direct import

cds.pde and cds.interpolate are lazily imported by the package root (PEP 562 module __getattr__). Accessing cds.pde imports the subpackage on first touch and caches it, so cds.pde and import cds.pde yield the very same module object while never putting a PDE solver on the import path of code that does not use one. Everything else in the table is imported eagerly by cds/__init__.py.

Quick Navigation

CDS is stable and actively developed. Contributions are welcome — see CONTRIBUTING.md.