This question is specifically about the robustness of mathematical models.
Special relativity can be derived from very basic principles. Assuming that space is homogeneous and isotropic and that relative velocity is measured the same by two observers is enough to show that the ransformation from one frame to another must have the form of a standard Lorentz transformation with some universal constant c, which is infinity in Galileian relativity.
Thus, the simple assumptions outlined above plus an estimate for a universal bound on speed are enough mathematically to derive all of special relativity. This shows that, in some sense, special relativity is the only theory that works in flat spacetime.
What about quantum mechanics? Dirac's book derives QM from general principles but throws in a few 'jumps' in logic that work great but seem to leave the door open to other mathematical theories. My question is,
Is there a small set of basic assumptions and experimental evidence which require quantum mechanics to have the mathematical form that it does?
For instance, does linear superposition of states (a simple assumption given experimental evidence) imply that observables must be linear operators (instead of other functions of states)? Is there any other inequivalent mathematical framework that satisfies linear superposition and the correspondence principle?