Proving a Mathematical hypothesis using Physics I've asked the question below on mathexchange here about 2 weeks ago. while I did not satisfied with the comments and answer there specially because the lack of examples and references that I was demanded for and also IMO it did not logically well discussed there, I've decided to put it as an inquiry on Physics community.
"We know that mathematical models of physical phenomena need to becomes more and more sophisticated as our observations become more precise and comprehensive by utilizing more advanced instruments that we upgrade over time. Also extending the mathematical models by means of mathematical tools and logical inferences lead us to predict and control physical incidents until observations confirm validity of the predictions. The problem arises where we couldn't find a comprehensive consistent mathematical model which fits over observations data and justify all of them. In such a case, the necessity of evolving mathematical theories seems crucial.
Questions here come to mind considering examples related to content above, imagine the observation data of Mercury’s movement that could not be justified by Newtonian mechanics achieved sooner than the appearance of non-Euclidean geometry, then wouldn’t it be influential on the advent of a different fifth postulate of geometry? Maybe the physics(or in more general term "the mother nature") can lead us for example to an axiom in order to make a mathematical hypothesis provable by adding it to our axiomatic system? How and under what circumstances can mathematics benefit from physics’ experimental achievements in order to evolve mathematical theories in a way that is not just be helpful for modelling physics but also useful for proving a mathematical hypotheses or altering a fundamental postulate of an axiomatic system?
Question : To what extent can advances in physics measurements and observations (not just ideas) aid mathematicians with proving a mathematical hypothesis that is based on an effectively generated and consistent axiomatic system like PA or ZFC?
For example consider dimensionless physical constant $\alpha$ “fine-structure constant” characterizing the strength of the electromagnetic interaction between elementary charged particles ,which Michael Atiyah has stated in his attempt to prove the Riemann hypothesis. Any precise explanation of this and other examples or references related to the topic are welcome."
 A: "Questions here come to mind considering examples related to content above, imagine the observation data of Mercury’s movement that could not be justified by Newtonian mechanics achieved sooner than the appearance of non-Euclidean geometry, then wouldn’t it be influential on the advent of a different fifth postulate of geometry?"
I am pretty sure that this was the case, this observation was known before Riemann and Einstein (though I cannot cite a reference it may be discussed in 300 years of Gravitation).  But why would the failure of Newtonian gravity and physics to describe Mercury's orbit lead to a questioning of Euclid's laws (if I understood your thesis correctly)?  This would not mean that the postulates of Euclidean geometry are wrong, merely that the laws of motion and gravity are misunderstood.  
On another note I do recall that techniques developed by physicists are sometimes useful in proving mathematical theorems or as computational tools.  An example of this is super symmetric quantum mechanics and the use of operator ordering to generate solutions to new ODE's that may have previously been unsolved.  Again, I cannot recall a specific citation but in graduate school I knew a group of theorists who were very good at this.  They used QM principles and SUSY QM to (1) find exact solutions to previously unsolved ODE's, and (2) use QM Postulates to develop proofs of theorems in ODE theory.  But I would not call this physics leading or inspiring mathematics as these results could have and eventually would have been found in math.    
