We want to make it easier and quicker to write well performing scientific simulations code that will (1) run fast on CPUs, and (2) be possible to maintain across the matrix of science experiments simulated vs. code versions which is inadvertently created in the course of working in an area. Instead of prototyping in a high-level language and then rewriting in a lower-level language, we created a DSL embedded in Python which is transparently usable, retains some OOP features, compiles to machine code, and executes at speed similar to C.