For a Hilbert space $\mathcal{H}_A$, I have seen the phrase

density matrices acting on $\mathcal{H}_A$

multiple times, e.g. here.

It is clear to me that if $\mathcal{H}_A$ has finite Hilbert dimension $n$, then this makes sense mathematically, because a density matrix $\rho$ can be written as $\rho \in \mathbb{C}^{n \times n}$ and elements $\phi$ of $\mathcal{H}_A$ can be written as $\phi \in \mathbb{C}^n$, so I can write down $\rho \phi \in \mathbb{C}^n$.

However, it is unclear what this means? The density matrix $\rho$ describes a (possibly mixed) state of a quantum system. But I can also interpret $\phi$ as a single state vector, describing a quantum system in a pure state.

So, what does $\rho\phi$ refer to (where $\rho \in \mathbb{C}^{n \times n}$ is a density matrix, and $\phi \in \mathbb{C}^n$ is an element of $\mathcal{H}_A$)? Can I interpret it? How can a density matrix (i.e., the representation of a state) act on states (on single state vectors)? Why do we interpret density matrices (which represent states) as operators?


It is common that one refers to a density matrix (or, equivalently, a density operator) $\rho$ as acting on a particular space $\mathcal{H}$. This serves to establish the "type" of $\rho$ in computer science parlance. In particular, when there are multiple spaces under consideration, it may be helpful for a reader to know that $\rho$ corresponds specifically to whatever abstract physical system is described by the space $\mathcal{H}$.

Referring to $\rho$ as acting on a space $\mathcal{H}$ also makes perfect sense, as the question points out, because $\rho$ can be viewed as a linear map from $\mathcal{H}$ to itself. The properties of $\rho$ that relate to its action as a linear mapping of this form are important and say a lot about the state. For example, the eigenvalues of $\rho$ describe the randomness or uncertainty inherent to that state.

However, this does not mean that if $\phi\in\mathcal{H}$ is a unit vector that describes a pure state, then $\rho\phi$ should have an interpretation. Such a vector might show up in a proof or calculation -- for example, the quantity $\langle \phi | \rho | \phi\rangle$, which is the inner product between $\phi$ and $\rho\phi$, is a commonly encountered quantity known as the fidelity (or squared fidelity) between the states represented by $\rho$ and $\phi$ -- but in my view the vector $\rho\phi$ is just a vector and does not have a natural or fundamental physical interpretation.


By "acting of $\mathcal{H}_A$" I believe you mean "acting on $\mathcal{H}_A$", which is what is written in the section you provided a link to.

What it means is "acting on a state in $\mathcal{H}_A$".

"I do not know what it means for a state to act on $\phi$."

You defined $\phi$ as a state. States do not act on states. Operators (such as $\rho$) act on states. The section you provided a link to, does not mention "acting on a state $\phi$" at any point.

  • $\begingroup$ Thanks for your answer! I have fixed the typo, and clarified the question. You pointed out exactly what I am confused about. There seem to be two kinds of states here, and one acts on the other... $\endgroup$ – Peter Jun 21 '18 at 7:41
  • $\begingroup$ You have completely changed the question, which makes my answer look silly. I think you should change it back and ask a new question "why are density matrices thought of as operators" $\endgroup$ – user1271772 Jun 21 '18 at 13:13

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.