S_4=1^4+2^4+...+n^4 (linear algebra method)
This problem provides a linear algebra method for computing sums of the form
with
positive integer.
1. Let
the real vector space of all polynomials of degree
. Consider the map
defined by
. Prove that
is a linear map.
2. Prove that

is constant.
3. Find the image by

of the elements of the canonical basis of
![\mathbb{R}_5[x]](http://latex.codecogs.com/png.latex?\mathbb{R}_5[x])
and determine
)
.
4. Using
3. , find an expression for

.
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
This is what I have for 1.
We have to prove that:
(1)
(2)
Proof (1):
![T(a(x)+b(x))=T((a+b)(x))=(a+b)(x+1)-(a+b)(x)=a(x+1)+b(x+1)-a(x)-b(x)=[a(x+1)-a(x)]+[b(x+1)-b(x)]=T(a(x))+T(b(x))](http://latex.codecogs.com/png.latex?T(a(x)+b(x))=T((a+b)(x))=(a+b)(x+1)-(a+b)(x)=a(x+1)+b(x+1)-a(x)-b(x)=[a(x+1)-a(x)]+[b(x+1)-b(x)]=T(a(x))+T(b(x)))
Proof (2):
![T(\lambda a(x))=\lambda a(x+1)-\lambda a(x)=\lambda[a(x+1)-a(x)]=\lambda T(a(x))](http://latex.codecogs.com/png.latex?T(\lambda a(x))=\lambda a(x+1)-\lambda a(x)=\lambda[a(x+1)-a(x)]=\lambda T(a(x)))
Therefore
is a linear map.
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
2. suppose T(p(x)) = 0, with
. then
) = a_0 + a_1(x+1) + a_2(x+1)^2 + a_3(x+1)^3 + a_4(x+1)^4 + a_5(x+1)^5\newline - a_0 - a_1x - a_2x^2 - a_3x^3 - a_4x^4 - a_5x^5)
 + (2a_2+3a_3+4a_4+5a_5)x + (3a_3+6a_4+10a_5)x^2\newline+(4a_4+10a_5)x^3+5a_5x ^5)
since this is by supposition the 0-polynomial, we have immediately that:

hence
, a constant polynomial. the reverse implication is trivial.
3.
(see above).
 = 1+x-x = 1)
(just take my word for it,ok?).
(think: truncated pascal's triangle).
(*whistling*)
.
so what is the pre-image of
? this amounts to solving:

so 




as we have seen T annihilates constants, so:
![T^{-1}(x) = \{c - \dfrac{1}{30}x + \dfrac{1}{3}x^3 - \dfrac{1}{2}x^4 + \dfrac{1}{5}x^5 \in \mathbb{R}_5[x] : c \in \mathbb{R}\}](http://latex.codecogs.com/png.latex?T^{-1}(x) = \{c - \dfrac{1}{30}x + \dfrac{1}{3}x^3 - \dfrac{1}{2}x^4 + \dfrac{1}{5}x^5 \in \mathbb{R}_5[x] : c \in \mathbb{R}\})
(short break for some etoufee for dinner)
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
ok, back. dinner was tasty!
4. let  = -\dfrac{1}{30}x + \dfrac{1}{3}x^3 - \dfrac{1}{2}x^4 + \dfrac{1}{5}x^5)
then, as we have seen,
.
this means that
for all real x. so, in particular:
 - q(1))
 - q(2))

 - q(n))
therefore,  - q(1))
but  - q(1) =)
 + \dfrac{1}{3}(n+1)^3 - \dfrac{1}{2}(n+1)^4 + \dfrac{1}{5}(n+1)^5 + \dfrac{1}{30} - \dfrac{1}{3} + \dfrac{1}{2} - \dfrac{1}{5})


hence 
which is the desired formula for
:)
(i'd like to add that this was a very entertaining problem, and that (in theory) Sn could be calculated from this method without much trouble, as the difficulty involved grows linearly with n. we still wind up with a upper triangular matrix for T, of rank n-1, so the resulting system of linear equations we get are easy to solve, although for large n, the repetition could get mind-numbing).
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
Beautiful! Can I ask, what is the source of the problem?
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
Quote:
Originally Posted by
Bourbaki
Beautiful! Can I ask, what is the source of the problem?
That is a problem proposed in an exam of Linear Algebra (Industrial Engineering School, UPM, Madrid)
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
(Off-topic) , there is a way to solve the general case using derivatives
Theorem
 = \sum^{n}_{k=0} k^p =\sum^{p+1}_{k=1}a_k n^k )
where
^{k-t}\;\; \mb{and}\;\; a_{p+1}=\frac{1}{p+1}. )
Proof
From
, we have
.
suppose = \sum^{p+1}_{k=1}a_k n^k )
with independent term
because
.
Apply the
-th derivative on
the result is
}(n) -s^{(p)}(n-1) =p! )
we have
then
} .)
We gonna find the other coefficients . We take the
-th derivative on
, with
, so
, taking
we have
}(0) =s^{t}(-1) )
Using
and applying again the
-th derivative
} (n)= \sum^{p+1}_{k=t}a_k(t!){ k \choose t} n^{k-t})
from
follows
{ k \choose t} 0^{k-t} = \sum^{p+1}_{k=t}a_k(t!){ k \choose t} (-1)^{k-t} \Rightarrow )
^{k-t} + a_t(t!)-a_{t+1}(t+1)! \Rightarrow)
^{k-t} )
Re: S_4=1^4+2^4+...+n^4 (linear algebra method)
and this should not be surprising, because:
1) the derivative is just a difference quotient. in fact, you don't need derivatives, you could do an induction argument on successive differences.
2) the rref of the matrix for T, and the differential operator D, are the same, so this is essentially the "same process".
3)pascal's triangle rulez (binomial coefficients are great with a little cream and sugar. oh wait, that's biscotti. my bad).