Use FrameX NDArray with NumPy ufuncs and reductions.

Tutorial: NumPy NDArray Interop

framex.NDArray supports NumPy protocols, so many NumPy operations work directly while retaining chunked storage.

Step 1: Create Chunked Arrays

import numpy as np
import framex as fx

x = fx.array(np.random.rand(2_000_000), chunks=250_000)
y = fx.array(np.random.rand(2_000_000), chunks=250_000)

Step 2: Run Ufuncs

z = np.sin(x) + np.log(y + 1.0)

Step 3: Use Reductions

print(np.mean(z))
print(np.max(z))
print(np.std(z))

Step 4: Use np.where

mask = z > 0.5
clipped = np.where(mask, z, 0.5)

Step 5: Convert Out

as_numpy = clipped.to_numpy()

When to Use NDArray

Use NDArray when your workflow is mostly numeric vector transforms and you want chunk-aware behavior with an easy NumPy-facing API.