darothen

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
SPHERE SCENE RAY TRACER

This is a basic port of the simple ray tracer from Chapter 6 of
"CUDA by Example", by Sanders and Kandrot. With a few exceptions
(notably, the `hit()` method is not bound to a struct containing 
sphere data and we use a numpy record array).

darothen

Sunburst visualization of modal importance aerosol activation experiment results.

darothen

darothen

darothen

darothen

def dataset_to_cube(ds, field):
    """ Construct an iris Cube from a field of a given
    xray DataSet. """

    raise NotImplementedError("`iris` deprecated for Python 3")

    dsf = ds[field] 

    ## Attach coordinates to the cube, using full dataset for lookup
    # dim_coords_and_dims = []

darothen

darothen

darothen

This is a simple example for using snakemake to automate a basic work pipeline.

Motivation

Makefiles and GNU Make are awesome for many reasons, and it's unforgivable for any scientist working with data processing pipelines to use them throughout their projects. But Makefiles, while feature-rich, are not really an ideal tool for automating complex data processing pipelines. If, by some chance, your analyses simply require you to collect different data, process them with identical procedures, collate them, and produce a plot, then sure - Makefiles will do. But in analyzing climate model output, I've found that I have to do a lot of quirky hacks to fit this sort of workflow model.

A perfect example is the analysis of hierarchical climate model output. It's quite common to run a climate model multiple times in a factorial factor, changing 2-3 parameters (say, an emissions dataset and a parameterization in the model). While you can pigeon-hole linear da

darothen