See https://pubs.acs.org/doi/abs/10.1021/acs.iecr.1c00847
Small computational speed bonus
(but maybe the compiler elides all the copies anyhow)
Some more testing is needed, but looks promising so far
Big speedup for gradient calculations
Switch critical tracing to use one of the molar concentrations as the fixed parameter when polishing Expose more critical methods in pybind11 interface
