kamodo 21.10.0

A functional api for scientific data

Kamodo

Kamodo is a CCMC tool for access, interpolation, and visualization of space weather models and data in python. Kamodo allows model developers to represent simulation results as mathematical functions which may be manipulated directly by end users. Kamodo handles unit conversion transparently and supports interactive science discovery through jupyter notebooks with minimal coding and is accessible through python.

The project page is located at the Community Coordinated Modeling Center, located at NASA Goddard Space Flight Center.

• Official site page https://ccmc.gsfc.nasa.gov/Kamodo/

Kamodo's official source code is hosted on github under a permissive NASA open source license:

• https://github.com/nasa/Kamodo

Periodic contributions to Kamodo are made from the unofficial repo located here

• https://github.com/asherp/Kamodo

Usage

Suppose we have a vector field defined by a function of positions in the x-y plane:

from kamodo import kamodofy
import numpy as np

x = np.linspace(-np.pi, np.pi, 25)
y = np.linspace(-np.pi, np.pi, 30)
xx, yy = np.meshgrid(x,y)
points = np.array(zip(xx.ravel(), yy.ravel()))

@kamodofy(units = 'km/s')
def fvec(rvec = points):
    ux = np.sin(rvec[:,0])
    uy = np.cos(rvec[:,1])
    return np.vstack((ux,uy)).T

The @kamodofy decorator lets us register this field with units to enable unit-conversion downstream:

from kamodo import Kamodo

kamodo = Kamodo(fvec = fvec)
kamodo

When run in a jupyter notebook, the above kamodo object will render as a set of equations:

$$\vec{f}{\left (\vec{r} \right )} [km/s] = \lambda{\left (\vec{r} \right )}$$

We can now evaluate our function using dot notation:

kamodo.fvec(np.array([[-1,1]]))

array([[-0.84147098,  0.54030231]])

We can perform unit conversion by function composition:

kamodo['gvec[m/s]'] = 'fvec'

kamodo automatically generates the appropriate multiplicative factors: $$\vec{g}{\left (\vec{r} \right )} [m/s] = 1000 \vec{f}{\left (\vec{r} \right )}$$ we can verify these results through evaluation

kamodo.gvec(np.array([[-1,1]]))

array([[-841.47098481,  540.30230587]])

Kamodo also generates quick-look graphics via function inspection.

import plotly.io as pio

fig = kamodo.plot('fvec')
pio.write_image(fig, 'images/fig2d-usage.svg')

Head over to the Introduction page for more details.

Getting started

Kamodo may be installed from pip

pip install kamodo

To get the latest version, install from Asher's fork:

pip install git+https://github.com/asherp/Kamodo.git

!!! note Asher's fork is periodically merged into the CCMC's official NASA version.

Kamodo Environment

We strongly recommend using the conda environment system to avoid library conflicts with your host machine's python.

Download and install miniconda from here. The advantage to using miniconda is that each new environment includes the bare-minimum for a project. This allows you to keep many different projects on a single work station.

Create Kamodo environment

Create a new environment for kamodo

conda create -n kamodo python==3.7
conda activate kamodo
(kamodo) pip install kamodo

!!! note The leading (kamodo) in your prompt indicates that you have activated the kamodo environment. From here on, anything you install will be isolated to the kamodo environment.

Loading example notebooks

If you want to run any of the notebooks in docs, you will need to install jupyter:

(kamodo) conda install jupyter

Navigate to the top-level of the kamodo repo, then point jupyter to docs/notebooks:

(kamodo) jupyter notebook docs/notebooks

This should open a browser window that will allow you to load any of the example notebooks.

Requirements

The following requirements are obtained by running pip install kamodo

• numpy
• scipy
• sympy
• pandas
• plotly==3.3
• pytest
• psutil
• conda install antlr-python-runtime (rendering latex)
• conda install -c plotly plotly-orca (for writing images)

!!! note plotly version in flux

Generating Docs

Kamodo's documentation site is a good example of how to embed your own plots in your own website. The documentation site is generated by the mkdocs package with some addons

• mkdocs - handles site generation and deployment (configured by top-level mkdocs.yaml)
• markdown-include - allows for embedding of markdown files (and graph divs) outside the docs folder
• python-markdown-math - enables LaTeX rendering
• mknotebooks - allows for the embedding of jupyter notebooks

All of the above requirements can be installed with this line:

pip install mkdocs python-markdown-math markdown-include mknotebooks

You can then generate the docs and serve locally with

mkdocs serve

To deploy your own documentation on github-pages:

mkdocs gh-deploy

This generates a gh-pages branch with the static site files and pushes it to github. Github automatically creates a website url based on that branch.