3 ms·
Would this be kind of the physics equivalent of nonstandard analysis in mathematics? Nonstandard analysis (NSA) is an approach to analysis that essentially make
by tzs 10y ago
Would this be kind of the physics equivalent of nonstandard analysis in mathematics? Nonstandard analysis (NSA) is an approach to analysis that essentially makes rigorous the vague notions of infinitesimals that Newton and Leibniz and the other pioneers of calculus based their work on.
Newton's fluxions and Leibniz's differentials provided great guidance, and in a lot of cases they even worked and gave correct results, but they were not really rigorously defined, and later mathematicians started running into serious problems with that approach. Those were resolved with the approach that we all now known and love (loath?): limits and the epsilon/delta approach to them.
NSA extends the real field to contain a new kind of number that is positive but is less than 1/N for any integer N, giving a new field called the hyperreals. You can then do things like define the derivative of f at x to be (f(x+h)-f(x))/h where h is an infinitesimal, work out that using ordinary algebraic manipulations, and then just drop any terms that are infinitesimal.
NSA goes far beyond just calculus. You can use these ideas all through analysis.
Seems pretty cool, and a lot clearer than a limits approach. So why has it not taken over?
I believe that a big part of it is that it has been shown that NSA and standard analysis are equivalent. Anything that you can prove in one can be proven in the other. Perhaps more important, one doesn't seem any better than the other at leading to deep insights. If you can't solve a problem in standard analysis, you probably aren't going to figure out the solution if you use NSA, and vice versa.
So while it might be overall nicer if we were starting from scratch to use NSA instead of standard analysis everywhere, so much is already done in standard analysis that switching now would be too painful.
- millstone 10y agoInteresting analogy but I'm not sure if it holds. Infinitesimals are an alternative to the limit-based formalism typically seen. It changes the way you do calculus. However Bohmian Mechanics is a different interpretation, not a different formalism. You can accept the interpretation, but you still end up solving the Schrödinger equation in the same way, using the same mathematical methods. The analogy works better for matrix mechanics as compared to Schrödinger's wave mechanics. These are mathematically quite different even though they give the same results.
- jayajay 10y agoThe Heisenberg picture and the Schrodinger picture the same, sometimes using one basis is more useful and more intuitive than using the other. Pilot Waves, on the other hand, are pretty much always superfluous. Please link me, if you have seen a situation where using Pilot Waves simplifies a calculation.
- naasking 10y agoThe pilot waves are always there whether or not you acknowledge them as such. There's nothing superfluous about them.
- ajkjk 10y agoI think you mean - you can always reparameterize things to have a 'pilot wave' around. That doesn't mean they're 'there', really. I would think that the 'default' interpretation would be excluding them, since (clearly) they're not necessary (evidence being: interpretations without them exist). That would be the definition of superfluous, right? Unless they give different experimental results somewhere, which I'm pretty sure in't the case.
- naasking 10y ago> I think you mean - you can always reparameterize things to have a 'pilot wave' around. That doesn't mean they're 'there', really. The wave equation is always there, or you wouldn't get quantum behaviour. There exist no-go theorems demonstrating that the wave function can't just be a reflection of our ignorance, it must be "ontic", ie. exist, in some real way. What's disputed is whether the particles need to be given separate existence from the waves that are necessarily there. > I would think that the 'default' interpretation would be excluding them, since (clearly) they're not necessary (evidence being: interpretations without them exist). That would be the definition of superfluous, right? No, because the interpretations you speak of can't actually explain the measurement problem without positing additional axioms which then make them less plausible. It's well known that pilot wave theories are more axiomatically parsimonious, meaning that they require fewer assumptions overall to explain all of our observations. For example, the Born rule must be assumed by most interpretations of QM, with little rhyme or reason other than we know it's empirically valid. But because pilot wave theories posit real physical entities with well understood properties, we can actually derive the Born rule. This is just one example that demonstrates how pilot wave theories positing additional real entities can make for an overall simpler set of assumptions.
- kkylin 10y agoI quite agree with millstone's comment below: interesting analogy, but not sure it is quite right here. And an off-topic, comment: I do agree with the sentiment that some things may be more easily expressed in nonstandard analysis. For an interesting experiment in that direction, see https://web.math.princeton.edu/~nelson/books/rept.pdf https://web.math.princeton.edu/~nelson/books/rept.pdf .
- qubex 10y agoThe most amazing aspect of Non-Standard Analysis is that the "tiny delta-x becomes h and you can then just take the limit by stripping out all the h symbols" approach was how I was first taught "Calculus" in International Baccalaureate (IB) Maths Higher in 1997-1999... an extraordinarily rickety foundation to build future intuition upon.
- thomasahle 10y agoNot to start a big discussion about infinitesimals, but don't they generalize a lot worse than limits? For example, what are their analogies in non-metric spaces? Say in topology or category theory?
- jostylr 10y agoA big difference is that standard analysis is well-defined on its own. The standard Copenhagen interpretation is simply not a well-defined theory because measurement is not well-defined, e.g., the cat problem. A more appropriate analogy to the SA v NSA would be GRW formulation vs Bohmian or even a well-defined many worlds (that does exist, but it does not involve splitting of worlds which is as problematic as collapse). These are different theories and they can lead to different generalizations. In particular, GRW seems more amenable to being relativistic without adding a foliation or some other structure. Bohmiam mechanics, on the other hand, does requires using a foliation though there are possibilities to use within the existing structures: https://arxiv.org/abs/1307.1714 https://arxiv.org/abs/1307.1714 I also think NSA has a naming problem. By its very name, how can it ever be standard? I think the issue is also that the limit fits in with numerical analysis in that it talks about errors. NSA sounds like it skips the error analysis and is only concerned with stuff at the limit which is certainly useful in many instances, but not always.
- MichaelBurge 10y ago> I also think NSA has a naming problem. By its very name, how can it ever be standard? "Nonstandard" isn't referring to how people view the field, but to additional elements you add to the model. Just to throw out a weird example, here's Edward Nelson's text on the subject: https://web.math.princeton.edu/~nelson/books/1.pdf https://web.math.princeton.edu/~nelson/books/1.pdf "Theorem 4. There is a finite set that contains every standard object." That is, in Nelson's system, the set of standard natural numbers is finite. That finite number is a nonstandard natural number, larger than any standard natural number. Mixing "nonstandard"(in the sense of an undefined predicate added to the base logic) with "nonstandard"(in the sense of some professors thinking it's weird) is a type error.
- libeclipse 10y agoIt should be noted that in mathematics, both of the analysis methods can be valid, while in physics, only one can be true. Someone correct me if I'm wrong here, but logically, standard theory and pilot theory can't BOTH be right, can they?