Finding formula for a Summation.

I'm given the sum of even squares to be $\frac{2n(n+1)(2n+1)}{3}$ e.g. 4+16+36+...

I'm also given the sum of odd squares to be $\frac{n(2n-1)(2n+1)}{3}$ 1+9+25+...

The problem is to find the summation formula of the series 1-4+9-16+25-... using the above facts.

I know I can write the sum of the even starting at 0 (e.g. 0+4+16+36+...) by $\frac{2(n-1)((n-1)+1)(2(n-1)+1)}{3}$ but I cannot figure out how to combine them to get the target series ( 1-4+9-16+25-...)

When just subtracting the evens starting from zero from the odds I always seem to get a sum that has more of the series than I want (1,6,15,28) when it should give (1,-3,6,-10) can anybody point me in the right direction?

-
What about separately considering the cases of n odd and n even? If n is odd, there's one more odd square than even. If n is even, there are equal numbers of odd and even squares. Though from the first few results, you might have a good guess already. – Mike Jan 11 '12 at 5:49
We have been given two gift expressions, one for the sum of the first $n$ even squares and one for the sum of the first $n$ odd squares. Throw away the second gift expression, divide the first gift expression by $4$. You get the formula for the sum of the first $n$ squares. (Presumably that's not what is intended, too simple!) – André Nicolas Jan 11 '12 at 6:55
First, please state when your sums begins and when it ends. It should just make the task easier to evaluate the sum. And similarly for 1-4+9-16 : do they both start from k=1 and stop at n ? And by the way your relation for the sum of the odds is only correct for n nonzero. – user20010 Jan 11 '12 at 9:48
@Ismail: The formulas are clearly intended to give the sum of the first $n$ positive even and odd squares, respectively, so they both work fine for $n=0$. – Brian M. Scott Jan 11 '12 at 10:33

Let $e_n$ be the sum of the squares of the first $n$ even positive integers, and let $o_n$ be the sum of the squares of the first $n$ odd positive integers; you have formulas for these. Let $s_n$ be the sum of the first $n$ terms of the target series, $1-4+9-16+-\dots$. Consider two cases, $n$ even and $n$ odd. (Note that I will be assuming throughout that $n$ is a positive integer.)

Suppose that $n$ is even, say $n=2m$. Then \begin{align*} s_n&=s_{2m}\\ &=o_m-e_m\\ &=\frac{m(2m-1)(2m+1)}3-\frac{2m(m+1)(2m+1)}3\\ &=\frac{m(2m+1)}3\Big((2m-1)-2(m+1)\Big)\\ &=-m(2m+1)\;. \end{align*}

Now suppose that $n$ is odd, say $n=2m+1$. Then \begin{align*} s_n&=s_{2m+1}\\ &=o_{m+1}-e_m\\ &=\frac{(m+1)(2m+1)(2m+3)}3-\frac{2m(m+1)(2m+1)}3\\ &=(m+1)(2m+1)\;. \end{align*}

Let’s run a quick spot-check. The first formula yields $s_4=-10$, which is correct, and the second yields $s_5=15$, which is also correct. All that remains is to combine them into a single nice expression, if possible. We have

$$s_n=\begin{cases} -m(2m+1),&\text{if }n=2m\\ (m+1)(2m+1),&\text{if }n=2m+1\;. \end{cases}$$

The algebraic sign is easily taken care of with a factor of $(-1)^{n+1}$, and $$\lceil n/2\rceil=\begin{cases} m,&\text{if }n=2m\\ m+1,&\text{if }n=2m+1\;, \end{cases}$$

so it only remains to handle the factor of $2m+1$. This can be done in many ways, but since we’re already using $\lceil n/2\rceil$, one easy way is to note that $2m+1=2\lceil n/2\rceil+(-1)^n$ no matter whether $n=2m$ or $n=2m+1$. Thus, we can finally write $$s_n=(-1)^{n+1}\left\lceil\frac{n}2\right\rceil\left(2\left\lceil\frac{n}2\right\rceil+(-1)^n\right)\;,$$ or, if you prefer, $$s_n=2(-1)^{n+1}\left\lceil\frac{n}2\right\rceil^2-\left\lceil\frac{n}2\right\rceil\;.$$

-