I have this determinant: $A=\begin{vmatrix} 1 & a_{1} & ... & a_{1}^{n-2} & a_{1}^{n} \\ 1 & a_{2} & ... & a_{2}^{n-2} & a_{2}^{n}\\ .& . & ... & . & .\\ .& . & ... & . & .\\ 1 & a_{n} & ... & a_{n}^{n-2} & a_{n}^{n} \end{vmatrix}$
Prove that: $A=\left ( a_{1}+a_{2}+...+a_{n} \right )\det\left ( B \right ),$ inside $B$ is the Vandermonde matrix.
Actually, I have not yet thought of a solution. Hope everyone helps me with the solution of this problem.
Here is an idea to prove the claim. It is based on the Gaussian elimination technique to compute the determinant of the standard Vandermonde matrix. There the first row is subtracted from all the others, then the factors $x_i-x_1$ are factored out. Finally, a column elimination is done to remove any lower order terms from the matrix columns. The remaining submatrix is again a Vandermonde matrix of lower order.
Now if we transfer this to the modified matrix in the question, then the last column of the transformed matrix after one step (subtracting first row from all others, factoring out $x_i-x_1$) consists of entries $x_i^{n-1} + x_i^{n-2}x_1$, $2\le i\le n$. (For the standard Vandermonde these entries would be of the form $x_i^{n-2}$.)
My claim is that after the $k$-step, the last column's entries are of the from $x_i^{n-k} + x_i^{n-k-1}\sum_{j=1}^kx_j$, $k+1\le i\le n$.
Thus after $n-1$ steps, the last remaining entry is $x_n + \sum_{j=1}^{n-1}x_j$ (instead of $1$ for the standard matrix). Now collecting all the factors, we find that the determinant is $\sum_{j=1}^{n}x_j$-times the determinant of the Vandermonde matrix.