reticulate | R Interface to Python | Machine Learning library

This isn't available through Conda, but is on PyPI. Try instead:

I think you only need to make a small change to your code. In particular, where you say:

Perhaps my failed attempts will help someone else solve this problem; my approach:

• boot up a clean CentOS 7 vm
• install R and some dependencies

I'm not sure if you can in R directly, but you definitely can in R Markdown. I use R Markdown to flip objects back and forth between the two.

I use a basic html_document YAML output. However, I don't typically knit this type of RMD, so I don't think it really matters what you put there if you use it the same way.

When you use reticulate you need an environment.

So first I'll have an R chunk:

I finally understood, there is a bug on CentOS 7: RcppTOML fails to compile with g++ 4.8.5 (20150623), but installs with g++ 5.3.1 (20160406) from scl-devtoolset-4.

The problem is that the R binary in rocker/r-ver:latest is compiled against a different BLAS library to the one which the numpy on PyPI is compiled against.

This was explained to me by Tomasz Kalinowski here.

The solution is to ensure numpy uses the same BLAS libraries as rocker/r-ver's R binary does. An easy way to ensure this is to compile numpy from source. This compilation could be performed at either image build-time or container runtime.

To compile numpy at container runtime we can leave our Dockerfile as is, and add a call to system2() after our initial call to reticulate::virtualenv_create(). Altering test.R to become:

Finally, in the function where I set up the python environment for the package, I called the R base system function where I pass the install command.

For example, for the pylearn-parsimony library which is not available on Conda, I used this line which worked:

Processes not started from a shell may not inherit environment variables from that shell. Start R in Terminal to make sure that your R process inherits PATH from the Terminal shell where you have run which. Something like

It is usually written wiki, where V* is the transpose of V:

And this is what you get back in scipy.linalg.svd:

Factorizes the matrix a into two unitary matrices U and Vh, and a 1-D array s of singular values (real, non-negative) such that a == U @ S @ Vh, where S is a suitably shaped matrix of zeros with main diagonal s.

Whereas for svd in R they return you V. Therefore should be: