Core¶

The top-level warpconvnet package re-exports the main subpackages and sets global configuration from environment variables at import time.

Subpackages¶

Package	Description
warpconvnet.geometry	Geometry containers (Points, Voxels, Grid, FactorGrid) and coordinate operations
warpconvnet.nn	Neural network modules and functional APIs for sparse/point convolutions
warpconvnet.ops	Low-level tensor operations (reductions, sampling)
warpconvnet.utils	Internal utilities (logging, caching, timing)
warpconvnet.constants	Environment-variable configuration and algorithm selection

Import example¶

import warpconvnet

# Geometry types
from warpconvnet.geometry.types.points import Points
from warpconvnet.geometry.types.voxels import Voxels

# Neural network modules
from warpconvnet.nn.modules.sparse_conv import SparseConv3d
from warpconvnet.nn.modules.point_conv import PointConv
from warpconvnet.nn.modules.sequential import Sequential

# Operations
from warpconvnet.ops.reductions import REDUCTIONS, row_reduction
from warpconvnet.ops.sampling import farthest_point_sampling