There are many experiments that prepare quantum systems in superpositions of different energy levels. For example, it is common to prepare cavities in coherent states, which are superpositions of SHO eigenstates.

One thing I know about physics is that energy is conserved. That means that if I start with a system in its ground state $E_0=0$ and want to evolve it to a superposition of states with energies $E_0$ and $E_1$, the additional energy must come from the environment, ie the rest of the universe. If before the evolution the rest of the universe had energy $E$, after the evolution I should be in a superposition of $E_0$ in the system and $E$ in the rest of the universe, and $E_1$ in the system and $E-E_1$ in the rest of the universe. Then the state of the system is entangled with the rest of the universe, and the system itself is in a mixed state rather than a superposition.

Clearly this reasoning is somehow incorrect, so what is the loophole that allows us to actually prepare superpositions instead of mixed states?

  • $\begingroup$ You might find this paper (arXiv) relevant reading in this regard. $\endgroup$ Mar 18 at 18:02
  • $\begingroup$ @fqq, well, I’m trying to make the system have a superposition of energy levels, but there is no way to do this without changing the energy of the environment, because energy is conserved. So it seems like that superposition is the “best” I can do! $\endgroup$ Mar 18 at 21:11

2 Answers 2


In most quantum optics experiments, states that are a superposition of energy states are generated using coherent states. For example the easiest way to prepare a cavity in a coherent state is to drive it with a laser, which outputs a coherent state. Coherent states have the property that removing a photon from the state does not change it, i.e., $a|\alpha\rangle \propto |\alpha\rangle$, so they can create energy superposition without creating entanglement.

But this begs the question of how to create coherent states, which are themselves superpositions of different energy states. It turns out that your reasoning is correct, and you need to bring measurement into the picture to get energy superposition. When your system is entangled with the rest of the universe like $|E_{universe}-E_0\rangle|E_0\rangle + |E_{universe}-E_1\rangle |E_1\rangle+...$, measurement of the system destroys the entanglement and can yield energy superpositions like $|E_0\rangle+e^{i\phi}|E_1\rangle+...$.

There's a very familiar example that illustrates this point nicely. Energy eigenstates of a free particle are plane waves $\psi_k(x) \propto e^{i k x}$, where the energy is just $\hbar^2 k^2/2m$. These are infinitely delocalized. We build up localized wave functions by taking coherent superpositions of them, like $\psi(x) = \int_{-\infty}^\infty \tilde{\psi}(k)\psi_k(x) dk$, where $\tilde{\psi}(k)$ is the Fourier transform of $\psi(x)$. So every position measurement generates coherence between energy eigenstates!

See this answer for more details on why realistic measurements tend to single out coherent states in systems that generate them (like lasers).


One intuitive argument could be that due to the size of the system, the "universe" state with energy $E$ and the universe state with energy level $E - E_1$ are more or less indistinguishable. Therefore a tensor product of $|E_0\rangle|E\rangle + |E_1\rangle|E-E_1\rangle$ is essentially the same as a pure state $(|E_0\rangle + |E_1\rangle)|universe\rangle$. Although theres probably a more rigorous argument that shows the same thing.

  • 2
    $\begingroup$ Although these states are difficult to distinguish in practice, there are still necessarily orthogonal. Energy is defined by a hermitian Hamiltonian, so eigenstates with different energies are always orthogonal. $\endgroup$
    – user34722
    Mar 18 at 18:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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