4
$\begingroup$

Let $A$ be an $n\times n$ matrix with entries $a_{i,j}$. Define an $(n-1)\times(n-1)$ matrix $B$ with entries $b_{i,j}=a_{1,1}a_{i+1,j+1}-a_{1,j+1}a_{i+1,1}$. Then $\det(B)=a_{1,1}^{n-2}\det(A)$.

I can prove this by direct computation, but it seems like something that may be well known or follow from other properties of determinants. Do you either know a reference for this result or know how to give a simple proof of it?

$\endgroup$
4
  • 1
    $\begingroup$ en.wikipedia.org/wiki/Dodgson_condensation $\endgroup$ Commented Jun 10, 2021 at 19:55
  • $\begingroup$ Ah, maybe it is not quite exactly the same as Dodgson condensation, but it seems very close... $\endgroup$ Commented Jun 10, 2021 at 19:57
  • $\begingroup$ I once needed a reference for an easy but obscure matrix identity and found it in Muir’s Treatise on the Theory of Determinants (I think Google Books has full text). These types of pursuits were much more fashionable back in the 19th century (as evidenced by the similarity to Dodgson condensation). Apologies I don’t have time at the moment to check if it’s in there. $\endgroup$
    – Erick Wong
    Commented Jun 10, 2021 at 23:58
  • $\begingroup$ cip.ifi.lmu.de/~grinberg/primes2015/kazh-exp.pdf gives a direct proof (although a longish one). $\endgroup$ Commented Jul 4, 2021 at 23:09

3 Answers 3

3
$\begingroup$

A simple proof is given in the Art of Problem Solving (it is entered as an "olympiade problem").

I reproduce the two-line proof for the record, with the change that $a_{11}\mapsto a_{nn}$:

Add the $n$-th row of $A$ to the $i$-th row, multiplied by $-a_{in}/a_{nn}$: $$\det{A}= \left| \begin{array}{cccccc} a_{1,1}-a_{n,1}\dfrac{a_{1,n}}{a_{n,n}} & \dots & a_{1,j}-a_{n,j}\dfrac{a_{1,n}}{a_{n,n}} & \dots & a_{1,n-1}-a_{n,n-1}\dfrac{a_{1,n}}{a_{n,n}} & 0 \\ \dots & \dots & \dots & \dots & \dots & \dots \\ a_{i,1}-a_{n,1}\dfrac{a_{i,n}}{a_{n,n}} & \dots & a_{i,j}-a_{n,j}\dfrac{a_{i,n}}{a_{n,n}} & \dots & a_{i,n-1}-a_{n,n-1}\dfrac{a_{i,n}}{a_{n,n}} & 0 \\ \dots & \dots & \dots & \dots & \dots & \dots \\ a_{n-1,1}-a_{n,1}\dfrac{a_{n-1,n}}{a_{n,n}} & \dots & a_{n-1,j}-a_{n,j}\dfrac{a_{n-1,n}}{a_{n,n}} & \dots & a_{n-1,n-1}-a_{n,n-1}\dfrac{a_{n-1,n}}{a_{n,n}} & 0 \\ a_{n,1} & \dots & a_{n,j} & \dots & a_{n,n-1} & a_{n,n} \end{array} \right|$$ $$=a_{n,n} \cdot \left| \begin{array}{ccccc} \dfrac{b_{1,1}}{a_{n,n}} & \dots & \dfrac{b_{1,j}}{a_{n,n}} & \dots & \dfrac{b_{1,n-1}}{a_{n,n}} \\ \dots & \dots & \dots & \dots & \dots \\ \dfrac{b_{i,1}}{a_{n,n}} & \dots & \dfrac{b_{i,j}}{a_{n,n}} & \dots & \dfrac{b_{i,n-1}}{a_{n,n}} \\ \dots & \dots & \dots & \dots & \dots \\ \dfrac{b_{n-1,1}}{a_{n,n}} & \dots & \dfrac{b_{n-1,j}}{a_{n,n}} & \dots & \dfrac{b_{n-1,n-1}}{a_{n,n}} \end{array} \right|= \dfrac{1}{{a_{n,n}}^{n-2}} \cdot \det{B} $$

$\endgroup$
5
  • 1
    $\begingroup$ You are probably tired answering such questions, but how did you find it? $\endgroup$ Commented Jun 10, 2021 at 20:18
  • 4
    $\begingroup$ this search returns the AoPS link at the bottom of the first results page. $\endgroup$ Commented Jun 10, 2021 at 20:20
  • 2
    $\begingroup$ million of thanks, I did not know about this engine $\endgroup$ Commented Jun 10, 2021 at 20:49
  • $\begingroup$ @FedorPetrov Thanks for calling attention to this search engine and to Carlo for linking it! $\endgroup$
    – Erick Wong
    Commented Jun 11, 2021 at 0:00
  • $\begingroup$ Note: This requires some reasoning to justify that $a_{n,n}$ can be assumed to be invertible. $\endgroup$ Commented Jul 4, 2021 at 23:07
4
$\begingroup$

This can also be done in terms of the Schur complement determinant formula.

The matrix $B$ is by definition $a_{1,1}A/{1}$, where $A/{1}$ is the Schur complement of $A$ with respect to the $1 \times 1$ submatrix with that one diagonal entry. The determinant formula for Schur complements says $$ \det(A)=a_{1,1}\det(A/{1})=\frac{\det(B)}{a_{1,1}^{n-2}}, $$ which is what we wanted.

$\endgroup$
1
$\begingroup$

Thank you to @carlo-beenakker, @erick-wong, @quizzical, and @sam-hopkins for helpful answers and comments.

Investigating further, I found that the statement in the question appears to be a well-known result of Chio from his 1853 manuscript Mémoire sur les fonctions connues sous le nom de résultantes ou de déterminans. I was unable to obtain Chio's manuscript but found a summary of it in Muir's 1911 work The Theory of Determinants in the Historical Order of Development, Volume II. Footnote 2 of this paper discusses the rendering of Chio's name. Sections 2 and 3 of this paper discuss the relationship between Chio's result and a more general result stated without proof by Sylvester in his 1851 article On the relation between the minor determinants of linearly equivalent quadratic functions.

I found an appropriate reference to a proof to be Eves's 1966 textbook Elementary Matrix Theory, which includes the statement in this question as Theorem 3.6.1.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .