10

My professor told us that in quantum mechanics a transformation is a symmetry transformation if $$ UH(\psi) = HU(\psi) $$

Can you give me an easy explanation for this definition?

Qmechanic
  • 201,751
SimoBartz
  • 1,790

2 Answers2

19

In a context like this, a symmetry is a transformation that converts solutions of the equation(s) of motion to other solutions of the equation(s) of motion.

In this case, the equation of motion is the Schrödinger equation $$ i\hbar\frac{d}{dt}\psi=H\psi. \tag{1} $$ We can multiply both sides of equation (1) by $U$ to get $$ Ui\hbar\frac{d}{dt}\psi=UH\psi. \tag{2} $$ If $UH=HU$ and $U$ is independent of time, then equation (2) may be rewritten as $$ i\hbar\frac{d}{dt}U\psi=HU\psi. \tag{3} $$ which says that if $\psi$ solves equation (1), then so does $U\psi$, so $U$ is a symmetry.

For a more general definition of symmetry in QM, see

Symmetry transformations on a quantum system; Definitions

Chiral Anomaly
  • 54,035
  • 3
    This is a good answer but it brings to another question, why do we call symmetry this condition? – SimoBartz Apr 08 '19 at 13:39
  • @SimoBartz That's a good question. In a more completely specified model, say with lots of local observables as in quantum field theory, we would require that a symmetry preserve things like the relationships between those observables in space and time. But in the present question, only the Hamiltonian is specified, so there is nothing else to preserve. – Chiral Anomaly Apr 08 '19 at 16:58
  • 1
    @SimoBartz, what does the word "symmetry" mean to you? Have you encountered it in other contexts, such as classical mechanics or geometry? – Vectornaut Apr 08 '19 at 21:48
  • @Vectornaut What if they answered yes to any of those? What would you say? – Krupip Apr 09 '19 at 17:01
  • Actually I'have never seen this concept before, my professor told us that when you have a symmetry transformation the system is invariant respect to that transformation. I imagine it means that nothing changes except the point of view. But if I transform a solution in another one maybe the new solution is completely different – SimoBartz Apr 10 '19 at 12:54
  • @SimoBartz The word "symmetry" is used with related-but-different meanings. So are the words "system is invariant." For example, we could define a symmetry (or invariance) to be any transformation that doesn't change the model's physical predictions. By that definition, every unitary transformation -- if applied both to the state-vectors and to all of the observables -- would trivially be a symmetry. The answer I gave here only offered a relatively easy motivation for the specific definition that you showed in the OP. – Chiral Anomaly Apr 11 '19 at 13:02
  • @SimoBartz And you're right: for most transformations $\psi\to U\psi$ that satisfy $UH=HU$, the states $\psi$ and $U\psi$ would give completely different predictions with respect to other observables. So defining a "symmetry" to be any $\psi\to U\psi$ for which $UH=HU$ is really only useful if $H$ is the only observable we care about. – Chiral Anomaly Apr 11 '19 at 13:02
1

What you have written there is nothing but the commutator. Consider for example the time evolution operator \begin{align*} U\left(t-t_{0}\right)=e^{-i\left(t-t_{0}\right) H} \end{align*} If $\psi\left(\xi_{1}, \dots, \xi_{N} ; t_{0}\right)$ is the wave function at time $t_0$ and $U(t−t0)$ is the time evolution operator that for all permutations $P$ satisfies $\left[U\left(t-t_{0}\right), P\right]=0$ then also $$\left(P U\left(t-t_{0}\right) \psi\right)\left(\xi_{1}, \ldots, \xi_{N} ; t_{0}\right)=\left(U\left(t-t_{0}\right) P \psi\right)\left(\xi_{1}, \ldots, \xi_{N} ; t_{0}\right)$$ This means that the permuted time evolved wave function is the same as the time evolved permuted wave function.

Another example would be if you consider identical particles. An arbitrary observable $A$ should be the same under the permutation operator $P$ if one has identical particles. This is to say: \begin{align*} [A, P]=0 \end{align*} for all $P\in S_N$ (in permutation group of $N$ particles).

Leviathan
  • 309