Assume all params are non-negative

[mstechly]

In many cases we can safely assume all the parameters in a routine are positive, which, in turn, might have two effects:

1. Will allow sympy to simplify the expressions more on the fly
2. Might be very useful for rewriters.

This can be easily done using sympy's nonnegative keyword: x=Symbol("x", nonnegative=True), and AFAIR is much easier if done at creation of the symbol, rather than modified later.
However, I'm not sure if we're not re-creating the symbols somewhere etc., so needs to be done with caution.

[dexter2206]

While we may certainly slap nonnegative keyword at symbol creation, for it to actually be useful and coherent some additional work would need to be done.

First, observe that sympy respects assumptions like nonnegative when doing simplification, but then doesn't check them e.g. when substituting. Consider

from sympy import Symbol, Max

a = Symbol("a", nonnegative=True)
expr = Max(0, a) # Simplifies to `a` given nonnegativity assumption
print(expr.subs({a: -500})) # -500, which is not equal Max(0, -500)

The example above is artificial, but you know what I mean - if we add any assumptions that are used for the simplification, we also should verify those assumptions are not violated, at least with trivial numeric substitutions (for general substitutions it might be hardly possible, but I wouldn't worry about it that much). Otherwise users might end up substituting into expressions values that are outside of their domain.

Secondly, we would need a way for specifying which symbols are nonnegative. Clearly, we cannot assume all of the symbols are noonegative (we wouldn't prohibit negative rotations angles, right?). This information would somehow need to be fed to the backend during expression parsing, because most of the input data is textual. This can be done, but would require some significant work. On top of that we would either have to alter QREF's schema, or abuse the meta field for storing assumptions.

That said, I'm not trying to discourage us from pursuing this enhancement, just pointing out that if we want to do this properly we would need to carefully think it through.