Euler's solution can be generalized, but those I have tried become useless rather fast.
Another solution of the Basel problem seems general enough to be possible to be applied to higher powers: the solution using Parseval's identity (theorem). Using Parseval's identity one can find the values of ζ for all even integers.
Why just don't we apply the same to s=3, and find odd powers as well?. The reason boils down to the fact that Lp is Hilbert iff p=2. The proof relies on the Parallelogram law in turn, which is founded on the notion of inner product space.
An inner product is defined for two vectors only [this is related to the two values of binary logic but here is not the place to pursue that].
Why? Sentimental reasons can be provided, such as `logic is binary', etc. but there is no logical necessity. One should in principle be able to define inner product for any number of vectors. In particular, for s=3 one can define ⟨⋅,⋅,⋅⟩:V×V×V→R by
⟨u,v,w⟩=∑iuiviwi.
One can create differential geometry accordingly:
ds3=gabcdxadxbdxc.
Next step, we should create an analogue Fourier analysis. Important problems arise here, such as `why don't we have triply-periodic functions' [not the usual answer], etc.
Fourier analysis is based on the Hilbert inner product of two functions
⟨f,g⟩=∫baf(x)ˉg(x) dx.
This requires two tools
- Generalization/analogue of complex conjugation, which in turn faces us with the old problem of `triplets',
- Inner product of three functions.
The second we already have, but for the first we must define a product
⋆:V×V×V→V,
such that our natural expectations are satisfied. We expect
- (x1,0,0)(x2,0,0)(x3,0,0)=(x1x2x3,0,0)
- (0,1,0)3=(−1,0,0)
- j3=0,j≠0⇒(0,0,1)3=(0,0,0)
We define basic trigonometric functions by the following system of ODEs
- ddzstz=ctz ttz
- ddzctz=−stz ttz
- ddzttz=stz ctz
No comments:
Post a Comment