# Meeting Notes III

- Consider a `conda` package for managing the xeus-cling dependencies
- Like Eigen
- Consider raw LAPACK
- ++ Xtensor is preferred ++
- [See here](https://xtensor.readthedocs.io/en/latest/numpy.html#linear-algebra)
> [name=Ondrej Will ask the Xtensor team for suggestions]

- Consider case studies
- Like a heat equation solver or some more exotic optimization study

- Manual compiled benchmarks
- [see here](https://github.com/certik/scipy-2016-symengine-talk/blob/master/talk.pdf)
- [Also this](https://meet.google.com/linkredirect?authuser=0&dest=https%3A%2F%2Fgithub.com%2Fcertik%2Fscipy-2016-symengine-talk%2Fblob%2Fmaster%2Fbenchmarks%2Fbenchmarks.ipynb)

- Simple examples for each binding
- More complete examples for Python and C++

- Consider the [ASDL represation](https://gitlab.com/lfortran/lfortran/-/blob/28482e86e811b1960138aac4c5c24ac2c1529a14/grammar/AST.asdl) instead of dealing with the current set of classes

## Previous Points

- [x] Can do the same docs as sympy for the simple cases