Colab does not make this easy. In principle all ingredients are there, since you can install Conda and then use the
cadabra2-jupyter-kernel package to install the Cadabra kernel. What's lacking is a simple way to switch the kernel of that running notebook; I haven't found anything so far.
An alternative is to use Binder https://mybinder.org/ , which allows you to setup an environment in a github repository and then start a notebook from that same repository. See
for an example of how to do this. The
environment.yml file in that repo specifies that we want to install the
cadabra2-jupyter-kernel conda package. If you go to the binder site and tell it to open a notebook from this repo, it will build a Docker image with this kernel installed, and allow you to open the
schwarzschild.ipynb notebook, asking you to specify which kernel you want to use.
Please let me know if you encounter problems with this, or if you figure out how to run the Cadabra kernel on Colab.