1

I've found a few determinants of this form and I would like to reach a simple expression for this determinant in terms of a and n. $$ D_{a,n}=\begin{vmatrix} 1 & a & a^2 &\cdots & a^{n-1} & a^n\\ a & 1 & a &\cdots & a^{n-2} & a^{n-1}\\ a^2 & a & 1 &\cdots & a^{n-3} & a^{n-2}\\ \vdots & \vdots & \vdots & \ddots &\vdots & \vdots \\ a^{n-1} & a^{n-2} & a^{n-3} & \cdots & 1 & a \\ a^n & a^{n-1} & a^{n-2} & \cdots & a & 1 \end{vmatrix}_{(n+1) \times (n+1)} $$

My idea was to subtract the first row multiplied by the leading coefficient (of the other rows) from all others. This will put $ (1-a^2)$ on the main diagonal if repeated n times, while also making $D_{a,n}$ the determinant of an upper triangular matrix.

I.e:

$$ D_{a,n}=\begin{vmatrix} 1 & a & a^2 &\cdots & a^{n-1} & a^n\\ 0 & 1-a^2 & a(1-a^2) &\cdots & a^{n-2}(1-a^2) & a^{n-1}(1-a^2)\\ 0 & a(1-a^2) & 1-a^4 &\cdots & a^{n-3}(1-a^4) & a^{n-2}(1-a^4)\\ \vdots & \vdots & \vdots & \ddots &\vdots & \vdots \\ 0 & a^{n-2}(1-a^2) & a^{n-3}(1-a^2) & \cdots & 1-a^{2(n-1)} & a(1-a^{2(n-1)}) \\ 0 & a^{n-1}(1-a^2) & a^{n-2}(1-a^2) & \cdots & a(1-a^{2(n-1)}) & 1-a^{2n} \end{vmatrix}_{(n+1) \times (n+1)} \\ $$

$$ D_{a,n}=\begin{vmatrix} 1 & a & a^2 &\cdots & a^{n-1} & a^n\\ 0 & 1-a^2 & a(1-a^2) &\cdots & a^{n-2}(1-a^2) & a^{n-1}(1-a^2)\\ 0 & 0 & 1-a^2 &\cdots & a^{n-3}(1-a^4) & a^{n-2}(1-a^4)\\ \vdots & \vdots & \vdots & \ddots &\vdots & \vdots \\ 0 & 0 & a^{n-3}(1-a^2)^2 & \cdots & 1-a^{2(n-2)} & a(1-a^{2(n-2)}) \\ 0 & 0 & a^{n-2}(1-a^2) & \cdots & a(1-a^{2(n-2)}) & 1-a^{2(n-1)} \end{vmatrix}_{(n+1) \times (n+1)} \\ \vdots $$

$$ D_{a,n}=\begin{vmatrix} 1 & a & a^2 &\cdots & a^{n-1} & a^n\\ 0 & 1-a^2 & a(1-a^2) &\cdots & a^{n-2}(1-a^2) & a^{n-1}(1-a^2)\\ 0 & 0 & 1-a^2 &\cdots & a^{n-3}(1-a^4) & a^{n-2}(1-a^4)\\ \vdots & \vdots & \vdots & \ddots &\vdots & \vdots \\ 0 & 0 & 0 & \cdots & 1-a^2 & something \\ 0 & 0 & 0 & \cdots & 0 & 1-a^2 \end{vmatrix}_{(n+1) \times (n+1)} $$ $$ \implies D_{a,n}= (1-a^2)^{n} $$ However, I don't believe this is rigorous enough proof to be used on an exam.

I'm hoping someone could point me in the right direction.

(I've thought of using induction, but it doesn't feel like the simplest approach)

(Also, what would "something" be if done the way I started?)

Edit: I just found out this is the determinant of a Toeplitz matrix, which has already been answered before here.

1 Answers1

0

Denote by $M$ the matrix in question (note that there are $n+1$ entries on the first row of $M$; thus $M$ is $(n+1)\times(n+1)$, not $n\times n$). Then $$ \pmatrix{1\\ -a&1\\ &\ddots&\ddots\\ &&-a&1}M=\pmatrix{1&\ast&\cdots&\ast\\ &1-a^2&\ddots&\vdots\\ &&\ddots&\ast\\ &&&1-a^2}. $$ Therefore $D_{a,n}=\det M=(1-a^2)^n$.

user1551
  • 149,263
  • Your point about the dimensions is great and I will edit the question accordingly. While I understand the values that are not on the main diagonal are unimportant since it will be an upper triangular matrix, I believe I can't simply say "the elements on the main diagonal will be 1-a^2 when the matrix is made upper triangular." Could you please help me to solve it more rigorously? – Nikola Petrovic Nov 30 '19 at 08:57
  • @NikolaPetrovic I would say it's rigorous enough, but if you want, you may express the equality in my answer as $LM=U$. Then $u_{ij}=m_{ij}-am_{i-1,j}$ for every $i>1$. When $i>j$, since $m_{ij}=am_{i-1,j}$ when $i>j$, we have $u_{ij}=0$. Therefore $U$ is an upper triangular matrix whose diagonal entries are given by $u_{11}=1$ and $u_{ii}=m_{ii}-am_{i-1,i}=1-a^2$ for every $i>1$. – user1551 Nov 30 '19 at 10:23
  • Thank you! I think that's a nice way of doing it. – Nikola Petrovic Nov 30 '19 at 12:17