Thursday, 17 January 2019

combinatorics - Can the Basel problem be solved by Leibniz today?



It is well known that Leibniz derived the series
π4=i=0(1)i2i+1,


but apparently he did not prove that
π26=i=11i2.

Euler did, in 1741 (unfortunately, after the demise of Leibniz). Note that this was also before the time of Fourier.



My question: do we now have the tools to prove (2) using solely (1) as the definition of π? Any positive/negative results would be much appreciated. Thanks!



Clarification: I am not looking for a full-fledged rigorous proof of (1)(2). An estimate that (2) should hold, given (1), would qualify as an answer.


Answer



Note that proving k=11k2=π26()

is equivalent to proving k=01(2k+1)2=π28()
The hope for proving () instead of () is that squaring π4 gives π216 and adding this twice gives us π28. We will in fact prove that k=1(2k+1)2=(k=(1)k(2k+1))2

Since we know k=0(1)k(2k+1)=π4
and Nk=0(1)k(2k+1)=1k=(N+1)(1)k(2k+1),

we have that k=(1)k(2k+1)=π2


Square the above to get
(k=Nk=N1(1)k(2k+1))2=(k=Nk=N1(1)k(2k+1))(j=Nj=N1(1)j(2j+1))=k=Nk=N1j=Nj=N1(1)j+k(2k+1)(2j+1)=k=Nj,k=N1jk(1)j+k2(jk)(1(2k+1)1(2j+1))+k=Nk=N1(1)2k(2k+1)(2k+1)

Hence,
(k=Nk=N1(1)k(2k+1))2k=Nk=N11(2k+1)(2k+1)=k=Nj,k=N1jk(1)j+k2(jk)(1(2k+1)1(2j+1))

Let us now show that
()k=Nj,k=N1jk(1)j+k2(jk)(1(2k+1)1(2j+1))0 as N


We have
()=k=Nj,k=N1jk(1)j+k2(jk)1(2k+1)k=Nj,k=N1jk(1)j+k2(jk)1(2j+1)=k=Nj,k=N1jk(1)j+k2(jk)1(2k+1)+k=Nj,k=N1jk(1)j+k2(kj)1(2j+1)=2×(k=Nj,k=N1jk(1)j+k2(jk)1(2k+1))=k=Nj,k=N1jk(1)j+k(jk)1(2k+1)=Nk=N1(1)k(2k+1)(Nj=N1jk(1)j(jk))(k)

Let us simplify (k) a bit. Assuming kN1, we have
Nj=N1jk(1)j(jk)=Nj=k+1(1)j(jk)+k1j=N1(1)j(jk)=((1)k+11+(1)k+22++(1)NNk)+((1)k1(1)+(1)k2(2)++(1)N1N1k)=(1)k+1N+|k|+1j=N|k|+1(1)jj

If k=N1, we have
Nj=N1jk(1)j(jk)=Nj=N(1)j(j+N+1)=(1)N12N+1j=1(1)jj

We now have
()=Nk=0(1)kk+(1)k1k12k+1=Nk=0(1)k(kk1)2k+1

Now for k0
(kk1)=(1)k+1N+k+1j=Nk+1(1)jj(1)kN+(k+1)+1j=N(k+1)+1(1)jj=2(1)k+1N+k+1j=Nk+1(1)jj+(1)N+1(1N+k+2+1Nk)

Hence,
|kk1|=O(1N)

|()|Nk=012k+1O(1/N)=O(log(2N+1)/N)0

Hence,
(k=Nk=N1(1)k(2k+1))2k=Nk=N11(2k+1)(2k+1)0

Hence,
(k=(1)k(2k+1))2=k=1(2k+1)(2k+1)=2k=01(2k+1)2=234ζ(2)


Hence,ζ(2)=23π24=π26


No comments:

Post a Comment

real analysis - How to find limhrightarrow0fracsin(ha)h

How to find limh0sin(ha)h without lhopital rule? I know when I use lhopital I easy get $$ \lim_{h\rightarrow 0}...