Imagine discrete orthonormal basis made of infinite set of kets $|\phi_1\rangle , ..., |\phi_n\rangle,...$ Completeness or closure of the basis is given by:

$\sum_{n=1}^{\infty} | \phi_n \rangle \langle \phi_n | = \hat {I}$ (From Zetilli's Quantum mechanics book)

So what I did manage is that the eigenvalues of the operator $|\phi_m\rangle\langle\phi_m|$ are all zeros except one which is the square of norm of $|\phi_m\rangle$. That means the matrix can be diagonalized but I don't really understand how you diagonalize matrices like this. Since the norm is 1, all entries should be zero except the $(n,n)$ one? Also doesn't diagonalizing a matrix change it? I'm a bit confused, please guide me.

  • $\begingroup$ Any state $|\psi >$ can be written as $\sum a_n |\phi_n>$, as the set is a basis. Now apply your operator and use orthonormality. $\endgroup$ Commented Nov 23, 2020 at 19:46
  • $\begingroup$ The kronecker Delta forces the summation to take only one value and we can take another summation from both sides to get our $\psi$ again. Thank you. $\endgroup$ Commented Nov 23, 2020 at 19:59

1 Answer 1


You can see that an operator of this type behaves as the identity in $\Bbb R^3$ with a basis $\{|i\rangle\}$:

$$\left(\sum_{i=1}^3|i\rangle\langle i|\right)|v\rangle=\sum_{i=1}^3|i\rangle\left(\langle i|v\rangle\right)=\sum_{i=1}^3|i\rangle v_i=\sum_{i=1}^3 v_i|i\rangle=|v\rangle \tag{1},$$

the generalisation to infinite dimensions means replacing the finite sum with an infinite one. In each term $\langle i|v\rangle$ gives the $i$th component of the vector $|v\rangle$. Going even further we can make a similar definition that occurs frequently in quantum mechanics:

$$\int dx\text{ }|x\rangle\langle x|\psi\rangle=\int dx\text{ }\psi(x)|x\rangle=|\psi\rangle. \tag{2}$$

In which $|x\rangle$ are the eigenstates of the position operator and $\psi(x)$ is the associated wavefunction of $|\psi\rangle$, therefore: $$\sum_{i=1}^3|i\rangle\langle i|=\Bbb I\quad \text{and}\quad \int dx\text{ }|x\rangle\langle x|=\Bbb I \tag{3},$$ in which $\Bbb I$ is the identity operator on the respective vector spaces.

  • $\begingroup$ Thanks. So we always define $|\psi(x)> = \int dx \psi(x) |x>$? I assume this is what we mean by wave function in position space? I think you just answered my second question too. Thank you. $\endgroup$ Commented Nov 23, 2020 at 20:13
  • $\begingroup$ Yes the value of the wavefunction at the point $x$ is essentially a component of the Hilbert space state vector $|\psi\rangle$. $\endgroup$
    – Charlie
    Commented Nov 23, 2020 at 20:14
  • $\begingroup$ Okay out of curiosity (and this is the last one, I promise!), For the particle in infinite square well we got $\psi(x) = \sqrt(2/a) Sin(nπx/a)$. How would you derive $|\psi>$ from this? I guess you'll need the |x> eigenstates in integration but what are they? If I use $\hat{R_x} |x> = X_x |x>$ eigenvalue equation I get nowhere because I don't know what the position operator is. I hope my questions make sense... $\endgroup$ Commented Nov 23, 2020 at 20:44
  • $\begingroup$ Well technically you would get $|\psi\rangle$ from $|\psi\rangle=\int dx\text{ }\psi(x)|x\rangle$, but we generally work with the wavefunctions $\psi(x)$ rather than the vectors $|\psi\rangle$ since it's easier to extract information from them. $\endgroup$
    – Charlie
    Commented Nov 23, 2020 at 21:25
  • $\begingroup$ @AbdulQadeer the expression $|\psi(x)\rangle$ does not make sense and it is wrong we have rather $\psi(x)=\langle\psi|x\rangle$ $\endgroup$
    – physshyp
    Commented Nov 23, 2020 at 21:49

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.