MedVision ad

Elegant Problem 2 (1 Viewer)

Archman

Member
Joined
Jul 29, 2003
Messages
337
Gender
Undisclosed
HSC
N/A
heres one, not exactly nice tho..

let sum be S
so S = Sum(nCa(a^2)) for a from 0 to n
2S = Sum(nCa(a^2)+nC(n-a)(n-a)^2)
2S = Sum( nCa(n^2-2an+2a^2))
2S = Sum(nCa * (n^2)) - Sum(nCa * 2a(n-a))
2S = 2^n * n^2 - n*(n-1)*2*Sum((n-2)C(a-1)) .... here, a is from 1 to n-1 as the terms of nC0*2*0*n and nCn*2*n*0 are both 0
2S = 2^n*n^2 - n*(n-1)*2*2^(n-2)
S=2^(n-1)*n^2-n*(n-1)*2^(n-2)
S=2^(n-2)*n*(2n-n+1)
S=2^(n-2)*n*(n+1)

hmm.. i would like to see a combinatorial solution tho..
 

underthesun

N1NJ4
Joined
Aug 10, 2002
Messages
1,781
Location
At the top of Riovanes Castle
Gender
Undisclosed
HSC
2010
First I considered the expansion of

(1 + x)<sup>n</sup> = <sup>n</sup>C<sub>0</sub> + <sup>n</sup>C<sub>1</sub>x + ... + <sup>n</sup>C<sub>n</sub>x<sup>n</sup>

I differentiated both side with respect to x, and got
n(1+x)<sup>n</sup> = <sup>n</sup>C<sub>1</sub> + 2<sup>n</sup>C<sub>2</sub>x + ... + n<sup>n</sup>C<sub>n</sub>x<sup>n-1</sup>

I multiplied both side by x, and got
nx(1+x)<sup>n</sup> = <sup>n</sup>C<sub>1</sub>x + 2<sup>n</sup>C<sub>2</sub>x<sup>2</sup> + ... + n<sup>n</sup>C<sub>n</sub>x<sup>n</sup>

then I differentiated both side :
LHS = <sup>d</sup>/<sub>dx</sub>nx(1+x)<sup>n</sup>

= n(1+x)<sup>n-1</sup> + nx(n-1)(1+x)<sup>n-2</sup>

now:

RHS = <sup>d</sup>/<sub>dx</sub> (<sup>n</sup>C<sub>1</sub>x + 2<sup>n</sup>C<sub>2</sub>x<sup>2</sup> + ... + n<sup>n</sup>C<sub>n</sub>x<sup>n</sup>)

= <sup>n</sup>C<sub>1</sub> + (2<sup>2</sup>)<sup>n</sup>C<sub>2</sub>x<sup></sup> + ... + (n<sup>2</sup>)<sup>n</sup>C<sub>n</sub>x<sup>n-1</sup>

and LHS = RHS still.

if we let x = 1, RHS is the expression at the beginning (ie the question). Hence, the sum of the expression = LHS when x =1

LHS simplifies to, when we sub x = 1 :

= (n<sup>2</sup> + n)(1+x)<sup>n-2</sup>

which is the required sum.

damn.. I actually went checking Tk / Tk+1 ratio for one hour (had no idea what to do) before I finally realised that to get the answer you have to do differentiation tricks..
 

Archman

Member
Joined
Jul 29, 2003
Messages
337
Gender
Undisclosed
HSC
N/A
ha, thats a lot nicer. Looks like calculus is sometimes useful in combinatorics too.
 

Archman

Member
Joined
Jul 29, 2003
Messages
337
Gender
Undisclosed
HSC
N/A
umm.. stuff that involves combinations and permutations etc.
 

underthesun

N1NJ4
Joined
Aug 10, 2002
Messages
1,781
Location
At the top of Riovanes Castle
Gender
Undisclosed
HSC
2010
oh ok.

btw so far, the only use of calculus in binomials is for proving (1 + x)<sup>n</sup> derivatives / integrals patterns, where you have 1nc1 + 2nc2 + 3nc3 or nc1 / 1 + nc2 / 2 etc..
 

OLDMAN

Member
Joined
Feb 20, 2003
Messages
251
Location
Mudgee
Gender
Undisclosed
HSC
N/A
underthesun: nice solid approach again.
Archman: don't self efface, a good solution displaying some fancy combination algebraic footwork.
 

Users Who Are Viewing This Thread (Users: 0, Guests: 1)

Top