Getting Started

Contents

Getting Started#

We’ll start with a simple example that runs locally.

Installation#

Install Cubed with pip:

pip install cubed

This installs a minimal set of dependencies for running Cubed, which is sufficient for the demo below. You can also install the diagnostics extra package, which is needed for later examples to provide things like progress bars and visualizations of the computation:

pip install "cubed[diagnostics]"

Alternatively, you can install Cubed with Conda (note that this doesn’t include the packages for diagnostics):

conda install -c conda-forge cubed

Demo#

First, we’ll create a small array a:

import cubed
import cubed.array_api as xp
spec = cubed.Spec(work_dir="tmp", allowed_mem="100kB")
a = xp.asarray([[1, 2, 3], [4, 5, 6], [7, 8, 9]], chunks=(2, 2), spec=spec)

Cubed implements the Python Array API standard, which is essentially a subset of NumPy, and is imported as xp by convention.

Notice that we also specify chunks, just like in Dask Array, and a Spec object that describes the resources available to run computations.

Next we create another array b and add to two array together to get c.

b = xp.asarray([[1, 1, 1], [1, 1, 1], [1, 1, 1]], chunks=(2, 2), spec=spec)
c = xp.add(a, b)

Cubed uses lazy evaluation, so nothing has been computed yet.

c.compute()

This runs the computation using the (default) local Python executor and prints the result (if run interactively):

array([[ 2,  3,  4],
       [ 5,  6,  7],
       [ 8,  9, 10]])

That’s it! For your next step you can read the user guide, have a look at configuration options, or see more examples to run locally or in the cloud.

Contents