Through the use of discrete mathematics, inefficient coding may be reduced that is more unavoidable in a smooth transition between numbers. Rather than assigning excess cpu and data. In substitute, sequence based recognition for non-visible calculations could be a path of least resistance where complex calculations are involved such as those involving pi. It would take longer with a new calculation, and shorter on familiar calculations similarly to muscle memory. This is already partially in place, for example loading into a game may take longer if the system has been deactivated for a prolonged duration than it would if you were launching it after having just exited it. Systematically chartering direct connections after indirect connections have been established a significant amount of times, especially as cpu is increasingly prioritized over data storage. To clarify, these discrete models would not be involved in new calculations nor in the user interface. I hypothesize that deep learning behaves bisimilarly, but that it could behave similarly under a more supervised construction of neural connections and pathways.
