When you begin configuring a compute capsule, you will be offered a number of
starter environments to choose between.

An environment consists of an operating system (typically Ubuntu Linux 16.04 or 18.04) and a minimal set of accompanying packages.

Some environments are pre-configured to support particular languages; these typically include a language interpreter, compiler, and/or framework. (Above, the 'Python 3.7.0' and the 'R 3.4.4' environments are pre-configured for specific languages.)

You can also start from a relatively blank slate (Ubuntu Linux 18.04, or 18.04 with GPU support), an environment without scientific programming languages pre-installed. This is a good choice for users with Linux programming experience.

All environments will have apt-get, a system level package manager, available by default. Language specific environments will have language-specific package managers available (e.g. CRAN or pip).

Hint: if you need multiple languages, or have a complex workflow, choose a Python environment (even if your code isn't, or mostly isn't, Python). As of February 2019, these environments have the Conda installer available, which is often the easiest way to install new languages through conda channels.

What's next? 

After selecting an environment, customize it by using the built-in package management system or the PostInstall script.

Switching between different base environments is also how to access a GPU.

Did this answer your question?