1
$\begingroup$

On page 32-33 of Dirac's book Principles of Quantum Mechanics, he considers the case when the linear, Hermitian$^1$ operator $\xi$ satisfies an algebraic equation

$$\phi(\xi)\equiv(\xi - c_1)(\xi - c_2)(\xi - c_3)...(\xi - c_n)=0\tag{18}$$

with the $c$'s being numbers, not assumed to be all different. Let the quotient when $\phi(\xi)$ is divided by $(\xi - c_r)$ be $\chi_r(\xi)$, so that

$$\phi(\xi)\equiv(\xi - c_r)\chi_r(\xi)\quad(r= 1,2,3,...,n)$$

After proving that the $c$'s are all different, he then says

Let $\chi_r(c_r)$ be the number obtained when $c_r$ is substituted for $\xi$ in the algebraic expression $\chi_r(\xi)$. Since the $c$'s are all different, $\chi_r(c_r)$ cannot vanish.

How is he able to conclude this?

--

$^1$ Note that a Hermitian operator is called a real operator in Dirac's non-standard terminology.

$\endgroup$
1
  • $\begingroup$ Because if $X_r(c_r)$ did vanish then it must be divisible by $(\xi-c_r)$ which would be a contradiction. $\endgroup$
    – lemon
    Commented Sep 24, 2014 at 18:26

2 Answers 2

2
$\begingroup$

If all the c's are different, then since $X_r(\xi)$ is the quotient when $\phi(\xi)$ is divided by $(\xi-c_r)$, $(\xi-c_r)$ cannot be a factor of $X_r(\xi)$ else two of the c's would equal $c_r$.

$X_r(c_r)$ is, by the remainder theorem (I imagine this has some other name), the remainder when $X_r(\xi)$ is divided by $(\xi-c_r)$. Since $(\xi-c_r)$ is not a factor of $X_r(\xi)$, $X_r(c_r)$ cannot be zero.

$\endgroup$
1
$\begingroup$

A very explicit argument:

$$\phi(\xi)=(\xi-c_1)\dots (\xi-c_{r-1})(\xi-c_r)(\xi-c_{r+1})\dots(\xi-c_n)$$

where $r\leq n$ and $c_k> c_l$ whenever $k>l$. We can order the $c_i$ like this because the $c_i$ are real.

Now, $$X_r(\xi)=(\xi-c_1)\dots (\xi-c_{r-1})(\xi-c_{r+1})\dots(\xi-c_n) $$

Clearly, the set of zeros of $X_r(\xi)$ is $\{c_i\ \bigl|\ i=1,\dots,r-1,r+1,\dots,n \}$. Note that this does not include $c_r$, so $X_r(c_r)$ does not vanish.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.