covertreex 0.4.0

Parallel compressed cover tree (PCCT) library targeting JAX-based Vecchia GP workloads.

Covertreex

High-performance cover tree library for k-NN queries, optimized for Vecchia-style Gaussian process pipelines.

Features

• Very fast, parallel implementation following Parallel Cover Trees and their Applications
• Hybrid Python/Numba + Rust implementation
• AVX2 SIMD optimized dot products in Rust backend
• Residual correlation metric with RBF and Matérn 5/2 kernels
• Hilbert curve ordering for cache-efficient tree construction

Installation

pip install covertreex

Quick Start

Basic Euclidean k-NN

import numpy as np
from covertreex import cover_tree

# Build tree and query
coords = np.random.randn(10000, 3).astype(np.float32)
tree = cover_tree(coords)
neighbors = tree.knn(k=10)

# With distances
neighbors, distances = tree.knn(k=10, return_distances=True)

Residual Correlation Metric (Vecchia GP)

For Gaussian process applications with Vecchia approximations:

import numpy as np
from covertreex import cover_tree
from covertreex.kernels import Matern52

coords = np.random.randn(10000, 3).astype(np.float32)

# Option 1: Provide a kernel (V-matrix built internally)
tree = cover_tree(coords, kernel=Matern52(lengthscale=1.0, variance=1.0))
neighbors = tree.knn(k=50)

# Option 2: Provide pre-computed V-matrix (from your GP)
# tree = cover_tree(coords, v_matrix=V, p_diag=p_diag)

# Predecessor constraint (for Vecchia): neighbor j must have j < query i
neighbors = tree.knn(k=50, predecessor_mode=True)

Engine Selection

from covertreex import cover_tree
from covertreex.kernels import Matern52

# cover_tree defaults to rust-hilbert (fastest)
tree = cover_tree(coords, kernel=Matern52(lengthscale=1.0), engine="rust-hilbert")
tree = cover_tree(coords, kernel=Matern52(lengthscale=1.0), engine="rust-natural")
tree = cover_tree(coords, kernel=Matern52(lengthscale=1.0), engine="python-numba")

Profile Presets

from covertreex import Runtime

# Load predefined configuration
runtime = Runtime.from_profile("residual-gold")

# With overrides
runtime = Runtime.from_profile("residual-gold", overrides=["seeds.global_seed=42"])

Available profiles: default, residual-gold, residual-fast, residual-audit, cpu-debug

Performance

Residual correlation k-NN benchmark (AMD Ryzen 9 9950X, N=32k, D=3, k=50):

Engine	Build Time	Query Throughput
python-numba	7.0s	40,000 q/s
rust-hilbert	0.9s	44,000 q/s

API Reference

cover_tree (recommended)

Factory function for building cover trees. Handles all configuration internally.

from covertreex import cover_tree
from covertreex.kernels import Matern52, RBF

# Euclidean distance (default)
tree = cover_tree(coords)

# Residual correlation with kernel
tree = cover_tree(coords, kernel=Matern52(lengthscale=1.0, variance=1.0))
tree = cover_tree(coords, kernel=RBF(lengthscale=2.0))

# Residual correlation with pre-computed V-matrix
tree = cover_tree(coords, v_matrix=V, p_diag=p_diag, kernel_diag=k_diag)

# Query
neighbors = tree.knn(k=10)
neighbors = tree.knn(k=50, predecessor_mode=True)  # Vecchia constraint
neighbors, distances = tree.knn(k=10, return_distances=True)

Kernel Classes

GP kernels for residual correlation metric:

from covertreex.kernels import Matern52, RBF

# Matérn 5/2 kernel (recommended for GP)
kernel = Matern52(lengthscale=1.0, variance=1.0)

# RBF (squared exponential) kernel
kernel = RBF(lengthscale=2.0, variance=1.0)

CoverTree (advanced)

Lower-level interface with explicit runtime configuration:

from covertreex import CoverTree, Runtime

runtime = Runtime(engine="rust-hilbert", metric="euclidean")
tree = CoverTree(runtime).fit(points)
neighbors = tree.knn(query_points, k=10)

Development

# Install in development mode
pip install -e ".[dev]"

# Build Rust backend
maturin develop --release

# Run tests
pytest

# Lint
ruff check covertreex

CLI (Testing)

A CLI is included for benchmarking and testing:

python -m cli.pcct --help
python -m cli.pcct query --engine rust-hilbert --tree-points 32768 --k 50

License

Apache 2.0 — See LICENSE for details.