Modules overview

cds2 ships forty-six importable modules plus a CLI. Every heavy computation is delegated to NumPy/SciPy; pandas carries tabular data; matplotlib renders figures.

Two thin wrappers, cds2.special and cds2.distributions, are deprecated since 4.3.0; prefer scipy.special and scipy.stats directly.

Core scientific stack

Module	Built on	Purpose
`cds2.linalg`	numpy.linalg	solve, det, inv, pinv, eig/eigh, SVD, least squares, cholesky, norms
`cds2.stats`	scipy.stats	t-tests, ANOVA, non-parametrics, correlations, chi-square, effect sizes
`cds2.optimize`	scipy.optimize	minimize, root finding, linear programming, curve fitting
`cds2.integrate`	scipy.integrate	quad, multi-dimensional integration, ODE solvers, Newton-Cotes rules
`cds2.interpolate`	scipy.interpolate	linear/cubic/pchip splines, lagrange, scattered and regular grids
`cds2.signals`	scipy.signal	FFT family, PSD/welch/spectrogram, Butterworth filters, peaks, envelope
`cds2.montecarlo`	numpy.random	seeded pi estimate, MC integration/expectation, Metropolis-Hastings
`cds2.graph`	scipy.sparse.csgraph	components, shortest paths, MST, PageRank, centrality measures
`cds2.ml`	numpy + cKDTree	linear/logistic regression, k-means++, PCA, kNN, metrics, generators
`cds2.timeseries`	pandas	moving average, EWM, decomposition, ACF/PACF, Ljung-Box
`cds2.viz`	matplotlib	series, histogram, scatter, heatmap, spectrum, regression, confusion
`cds2.io`	pandas	CSV/JSON read-write, optional Excel/Parquet bridges, DataFrame summaries
`cds2.calculus`	NumPy/SciPy	derivative, complex-step gradient, jacobian, hessian, error propagation
`cds2.special` (deprecated)	scipy.special	gamma, erf, Bessels, elliptics, orthogonal polynomials, zeta
`cds2.sparse`	scipy.sparse.linalg	CG/GMRES/BiCGSTAB, Lanczos eigenpairs, truncated SVD
`cds2.spectral`	scipy.sparse	Laplacians, Fiedler vector, algebraic connectivity, spectral clustering
`cds2.distributions` (deprecated)	scipy.stats	pdf/cdf/ppf for twenty-plus distributions

Discovery and modelling

Module	Built on	Purpose
`cds2.infotheory`	NumPy	entropy, KL / Jensen-Shannon divergence, mutual information, permutation entropy
`cds2.chaos`	NumPy	delay embedding, Lyapunov exponents, correlation dimension, Hurst exponent
`cds2.bayes`	scipy.stats	conjugate updates, credible intervals, naive Bayes, Metropolis posteriors
`cds2.metaheuristics`	NumPy	genetic algorithm, particle swarm optimization, simulated annealing
`cds2.geometry`	scipy.spatial	convex hull, closest pair, polygons, point-in-polygon, intersections
`cds2.rl`	NumPy	bandits (epsilon-greedy, UCB1), tabular Q-learning, grid world
`cds2.quality`	NumPy + scipy.stats	Shewhart/EWMA/CUSUM/p control charts, Cp/Cpk capability, defective PPM
`cds2.design`	NumPy	full & fractional factorial DOE, Latin hypercube sampling, central composite
`cds2.modeling`	pure Python	expression trees, symbolic diff/integral, polynomial solving, MathModel
`cds2.hypothesis`	pure Python	heuristic hypothesis generation with confidence scores
`cds2.knowledge`	pure Python	knowledge graph, notebook, ranked search
`cds2.scientific`	pure Python	physical constants, formulas and unit conversion
`cds2.quantum`	NumPy	statevector circuit simulator up to 16 qubits
`cds2.nlp`	NumPy	scalar autograd, BPE tokenizer, attention, mini-GPT forward pass
`cds2.cli`	argparse	`cds2` console entry point

Import styles

Flat re-exports live on the package root for the most common helpers:

import cds2

cds2.pagerank
cds2.pi_estimate
cds2.minimize

Each module can also be imported directly:

from cds2.graph import pagerank

Versioning

Static semantic versioning: cds2.__version__ mirrors the git tag. The v1.x zero-dependency line is maintained separately in its own repository.