Stella

"Stella has a relatively narrow focus: Python programs with a medium size core that needs to be executed very efficiently. Simulations are a prime example: In particular stochastic simulations need to be sampled many times to get relevant results which are statistically relevant, but even regular simulations often are extremely slow when run in the regular Python interpreter.
The key features of Stella are:

• compiled with LLVM for execution times similar to C programs
• statically typed; ensured correctness for the computationally expensive parts
• no interaction with the Python run-time to avoid any unintended performance penalties
• completely valid Python code for verification and debugging purposes
• easy integration with C libraries with the use of Cython
• excellent functional test coverage
• compatible with numpy arrays

Note: Stella is currently in an alpha stage: it is a research project, albeit a well tested one. But there is no guarantee that the program you want to run will be completely supported."

https://github.com/squisher/stella

https://digamma.cs.unm.edu/~dmohr/stella/dsldi-dmohr.pdf

https://verc.enes.org/ISENES2/.../Talks/.../osuna_stella-perspecive-of-the-big-picture