I understand from the standard teleportation protocol that 1 ebit is used up in teleporting 1 qubit and thus, cannot be used again -- and thus, we need 1 fresh ebit of shared entanglement between Alice and Bob per each qubit that Alice wants to teleport to Bob. Furthermore, the teleported state could have a phase-flip or a bit-flip error and thus, the $2$ bit syndrome of Alice's measurements need to be sent to Bob for Bob to recover the correct qubit.

However, is there a general proof of this statement apart from the heuristic argument? If so, is there a fundamental no-go theorem of quantum mechanics behind this restriction? For example, the fact that you need $2$ classical bits to be sent in order to teleport $1$ qubit appears to be less than optimal, i.e., even if it were possible to teleport a qubit using only $1$ classical bit (along with the $1$ ebit, of course), the no-communication theorem wouldn't be violated because one can only extract $1$ bit of classical information from $1$ qubit. Is there some other basic tenet of quantum mechanics that would be violated if this were to be possible?

  • $\begingroup$ Hmm, one thing I can think of is that if the qubit being teleported is entangled to another qubit that Bob has then superdense coding could be used to send $2$ bits of information and thus, if the teleportation only required $1$ bit of communication then we'd be in trouble vis-a-vis no-communication theorem. $\endgroup$ Jul 30, 2022 at 16:35

1 Answer 1


To prove that teleportation of a single qubit cannot be accomplished with less than 2 classical bits, you can assume the opposite to show that No-Signalling was violated (as your comment suggests).

Recall the two protocols that we know we can perform:

  1. TP (Teleportation): Alice transmits $1\text{ qubit}$ to Bob by physically sending $2\text{ cbits}$, consuming $1 \text{ ebit}$:

$$ \text{TP:}\qquad1\text{ ebit} + 2 \text{ cbits} \rightarrow 1 \text{ qubit}\tag{1} $$

  1. SD (Superdense Coding): Alice transmits $2\text{ cbits}$ to Bob by physically sending $1\text{ qubit}$ and consuming $1 \text{ ebit}$ in the process: $$ \text{SD:}\qquad1\text{ ebit} + 1 \text{ qubit} \rightarrow 2 \text{ cbits} \tag{2} $$

Now suppose there is a version of Teleportation (call it SuperTP) that transmits one qubit to bob but sends $x < 2$ cbits. This allows us to do the following: We will perform SD, but when it comes time to physically send 1 qubit, we will instead perform SuperTP to transmit that single qubit worth of information using $x <2$ cbits:

$$ \text{SuperTP:}\qquad1\text{ ebit} + x \text{ cbits} \rightarrow 1 \text{ qubit}\tag{3} $$

Substituting SuperTP for the physically transmitted qubit on the LHS of Eq. (2), we can describe a composition of the SuperTP and SD protocols as

$$ \text{SD[SuperTP]:}\qquad2\text{ ebits} + x \text{ cbits} \rightarrow 2 \text{ cbits} \tag{4} $$

Now it is clear that any $x<2$ will violate the No-Signalling principle, and so SuperTP cannot exist. In this sense, Teleportation with cbits and qubits is optimal; it uses the lower bound of $x=2$ cbits that does not violate No-Signalling. Furthermore, we have not assumed anything about the optimality of SD for this proof: A more efficient protocol for SD would only drive the lower bound for $x$ up. Yet, the existence of TP in Eq. (1) allows us to do this exact same procedure to show that SD is optimal, by creating a composition TP[SuperSD] (you might find it worthwhile to work through that). In this sense, there is a duality between SD and TP where each one's existence implies the other's optimality.

  • 1
    $\begingroup$ Thanks for your answer! Can you also comment on the optimality of the number ebits? Again, heuristically, I get it why we need 1 ebit for 1 qubit to be teleported (because any teleportation scheme would need a "phase kickback" step that would require Alice to make measurements and that would use up the entanglement) but I don't know what's the proper formal argument. Thanks again :) $\endgroup$ Jul 30, 2022 at 18:46

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.