Overview

What is BCE?

BCE stands for the Berkeley Common (or Compute, or Collaborative…) Environment. It is designed to provide a common Linux computational environment for classwork and research.

We’ve presented on BCE at SciPy2014. Here are slides and a draft of the conference paper.

BCE Vision

The goal for the BCE is to provide both the ready-made environments and also the “recipes” or scripts setting up these environments. It should be easy for a competent Linux user to create recipes for custom tools that might not be broadly useful (and thus, not already in BCE).

BCE is designed for classwork and research in the sciences at Berkeley, broadly defined to include social science, life science, physical science, and engineering. Using these tools, users can start up a virtual machine (VM) with a standardized Linux operating environment containing a set of standard software for scientific computing. The user can start the VM on their laptop, on a university server, or in the cloud. Furthermore, users will be able to modify the instructions for producing or modifying the virtual machine in a reproducible way for communication with and distribution to others.

We envision the following core use cases:

  • creating a common computing environment for a course or workshop,
  • creating a common computational environment to be shared by a group of researchers or students, and
  • disseminating the computational environment so outsiders can reproduce the results of a group.

What problems does BCE solve for you?

  • No more obscure installation issues - download and run a single virtual machine or get the same environment on a bare metal or virtual server.
  • I’m teaching a class - when you tell a student that a program behaves a certain way, it does!
  • I’m collaborating on some scientific research - now all of your collaborators can run your code without complex installation instructions.

To accomplish this, we envision that BCE will encompass the following:

  • a reproducible workflow that creates the standard VM/image with standard scientific computing software such as Python, R, git, etc.,
  • a standard binary image, produced by the workflow, that can be distributed as is and used on-the-fly with VirtualBox or VMWare Player with minimal dependencies, and
  • (possibly) an augmented workflow that represents multiple possible distributions tailored for different types of uses (e.g., different disciplines, different computational needs, class vs. research use, etc.). This might represent either a sequence or a tree of possible VMs.

Why is the URL “Collaboratool?”

“Collaboratool” was conceived as a project for building, integrating, and deploying tools that support portable, reproducible data science. The project has since been rebranded as BCE, but we haven’t updated the URL.

Who are we?

BCE is a project that started in the D-Lab, with collaboration from Computer Science (EECS), the Statistical Computing Facility (SCF), Berkeley Research Computing (BRC), and the School of Information.