---
title: 'Port-Based Teleportation: Theory & Applications'
url: https://www.emergentmind.com/topics/port-based-teleportation
type: topic
---

# Port-Based Teleportation: Theory & Applications

Port-based teleportation (PBT) is a family of quantum teleportation protocols in which Alice’s classical message identifies **which** one of Bob’s \(N\) subsystems contains the output state, so Bob performs only port selection and discarding rather than an outcome-dependent correction unitary. In the standard setting, Alice and Bob share \(N\) bipartite \(d\)-dimensional entangled pairs, Alice performs a joint measurement on the input and her halves of the ports, and the protocol is evaluated either by success probability or by entanglement fidelity, depending on whether a failure branch is allowed. This correction-free decoding makes PBT structurally different from Bennett teleportation, but perfect deterministic PBT is impossible with finitely many ports; its importance comes instead from exact finite-\(N\) characterizations, asymptotic optimality theory, and its role in programmable processing, channel simulation, and asynchronous or instantaneous non-local tasks [1612.09260][1809.10751].

## 1. Operational definition and protocol variants

In deterministic PBT, Alice and Bob share \(N\) ports \(A_1B_1,\dots,A_NB_N\), each local system is \(d\)-dimensional, and the shared resource can be written as
\[
|\Psi\rangle_{AB}=(O_A\otimes \mathbf{1}_B)\bigotimes_{i=1}^N |\psi_d^+\rangle_{A_iB_i},
\qquad
|\psi_d^+\rangle=\frac{1}{\sqrt d}\sum_{i=1}^d |ii\rangle,
\]
with \(\operatorname{Tr}(O_A^\dagger O_A)=d^N\). Alice measures the input together with \(A_1,\dots,A_N\), sends an outcome \(a\in\{1,\dots,N\}\), and Bob keeps \(B_a\). In the non-optimized scheme \(O_A=\mathbf{1}_A\); in the optimized scheme \(O_A\) is chosen to maximize teleportation fidelity [2105.14886].

The two standard variants are probabilistic and deterministic PBT. Probabilistic PBT has a failure outcome \(0\); when the outcome is \(a\ge 1\), teleportation is perfect, and the figure of merit is the success probability. Deterministic PBT has only \(N\) output labels, so some port is always produced, but the induced channel is approximate and is quantified by entanglement fidelity \(F\), with average state fidelity
\[
f=\frac{Fd+1}{d+1}.
\]
Operationally, probabilistic PBT optimizes success subject to exact conditional transmission, whereas deterministic PBT optimizes fidelity under guaranteed output [1612.09260].

A recurrent misconception is that “no correction” means “no classical communication.” PBT still requires one-way classical communication from Alice to Bob; what disappears is the need for Bob to implement a nontrivial unitary conditioned on that message. The message only selects a port. This distinction is precisely what makes PBT composable in settings where intermediate adaptive correction unitaries are undesirable or impossible [1809.10751].

## 2. Representation-theoretic structure

The exact analysis of PBT is controlled by the spectrum of the PBT operator
\[
\rho=\sum_{a=1}^N \varrho_a
      =\frac{1}{d^N}\sum_{a=1}^{N}V^{t_C}_{(CA_a)}\otimes \mathbf{1}_{\overline{A_a}},
\]
which belongs to the algebra of partially transposed permutation operators. This observation converts the apparent \(d^{N+1}\)-dimensional spectral problem into a structured problem in the representation theory of \(S_n\), \(U(d)\), and the partially transposed permutation algebra [1612.09260].

Schur–Weyl duality and the partially reduced irreducible representation formalism organize the relevant blocks by pairs \((\alpha,\mu)\), where \(\alpha\vdash N-1\) and \(\mu\vdash N\) is obtained from \(\alpha\) by adding one box. In this block basis, the eigenvalues of \(\rho\) are explicit:
\[
\lambda_\mu(\alpha)=\frac{N}{d^N}\frac{m_\mu d_\alpha}{m_\alpha d_\mu},
\]
with \(d_\alpha,d_\mu\) the dimensions of the symmetric-group irreps and \(m_\alpha,m_\mu\) the corresponding Schur–Weyl multiplicities. This is the core exact finite-\(N\) structural result underlying both deterministic and probabilistic PBT in arbitrary local dimension [1612.09260].

The same machinery also controls square-root measurements and their square roots. In the recycling analysis, the relevant SRM blocks are either genuine projectors or pseudo-projectors depending on whether a forbidden irrep \(\theta\) of height \(d+1\) appears. That distinction determines the explicit blockwise square root and makes one-round recycling fidelities computable in arbitrary dimension by purely group-theoretic data [2105.14886].

A practical consequence is algorithmic. For fixed \(d\), the number of Young diagrams with bounded height grows polynomially in \(N\), so the properties of any fixed-dimension PBT scheme can be determined in polynomial time once the representation-theoretic data are available [1612.09260].

## 3. Performance regimes and asymptotics

For probabilistic PBT with optimization over both measurement and resource state, the optimal success probability is exactly
\[
p_{\mathrm{opt}}=\frac{N}{N+d^2-1}.
\]
This formula is valid for all \(d\) and \(N\), so for fixed \(d\), \(p_{\mathrm{opt}}\to 1\) as \(N\to\infty\), while for fixed \(N\), \(p_{\mathrm{opt}}\to 0\) as \(d\to\infty\) [1612.09260].

Deterministic PBT exhibits a sharper separation between EPR-restricted and fully optimized resource states. With maximally entangled ports and the standard pretty-good measurement, the entanglement fidelity obeys
\[
F_d^{\mathrm{std}}(N)=1-\frac{d^2-1}{4N}+O(N^{-3/2+\delta})
\]
for every \(\delta>0\). By contrast, in fully optimized deterministic PBT,
\[
F_d^*(N)=1-\Theta(N^{-2})
\qquad (d\ \text{fixed},\ N\to\infty).
\]
Thus optimization of the resource state changes the asymptotic rate from \(O(N^{-1})\) to \(O(N^{-2})\), not merely the constant factor [1809.10751].

The converse side is equally important. The optimal deterministic error is fundamentally quadratic in \(1/N\): achievability is governed by the first Dirichlet eigenvalue of the Laplacian on the ordered simplex, while converse bounds match the same \(N^{-2}\) order. This identifies the asymptotic cost of removing Bob’s correction operation in deterministic teleportation [1809.10751].

A different regime emerges at fixed low \(N\) and large local dimension. For PGM-based PBT with \(N=2,3,4\), exact higher-dimensional calculations show that the entanglement fidelity satisfies
\[
F_e\le \frac{N}{d^2},
\qquad
F_e\ge \frac{N}{d^2+N-1},
\]
and, for fixed \(N\),
\[
F_e\sim \frac{N}{d^2}\qquad (d\to\infty).
\]
This indicates that high-dimensional PBT with few ports becomes fidelity-limited by \(d^{-2}\) scaling even though the port-identification measurement itself becomes nearly perfect [2207.04593].

## 4. Optimal measurements and optimized resources

A central structural result is that the pretty-good measurement, or square-root measurement,
\[
E_i=\bar{\sigma}^{-1/2}\,p_i\sigma_i\,\bar{\sigma}^{-1/2},
\]
is not merely near-optimal in deterministic PBT: it is exactly optimal for the standard port state of \(N\) maximally entangled pairs, and the **same specific measurement** remains optimal even when the port state is itself optimized over all admissible symmetric choices. In other words, within deterministic PBT, optimization changes the resource state but not the optimal measurement [2008.11194].

The proof strategy is based on semidefinite-program duality plus symmetry reduction. PBT is recast as a structured state-discrimination problem for the ensemble of reduced resource states \(\{\sigma^i\}\), and the dual-feasible operator constructed from the PGM performance matches the primal value exactly. The representation-theoretic block structure supplies the explicit dual certificate [2008.11194].

In the fully optimized deterministic setting, the remaining optimization can be expressed through the teleportation matrix \(M_F\), a combinatorial object indexed by Young diagrams of \(S(N)\). The optimal fidelity is
\[
F_{\mathrm{opt}}=\frac{1}{d^2}\|M_F^d\|_\infty,
\]
where \(M_F^d\) is the principal submatrix corresponding to diagrams of height at most \(d\). When \(d\ge N\), the full spectrum is explicit,
\[
\operatorname{spec}(M_F)=\{0,1,2,\ldots,N-2,N\},
\]
so \(\|M_F\|_\infty=N\) and \(F_{\mathrm{opt}}=N/d^2\) in that regime [1707.08456].

For qubits, this specializes to the exact formula
\[
F_{\mathrm{opt}}=\cos^2\!\left(\frac{\pi}{N+2}\right),
\]
recovering the original optimal qubit result in closed form. The optimized resource operator is diagonal in the Schur–Weyl decomposition and is determined by the Perron eigenvector of \(M_F^d\) [1707.08456].

## 5. Recycling, degradation, and noisy resources

PBT consumes entanglement nontrivially: after one teleportation round, the unused ports are no longer a tensor product of maximally entangled pairs. This motivates recycling protocols, where the post-measurement resource is reused. For deterministic PBT in arbitrary dimension, one-round recycling fidelities were derived explicitly in terms of Schur–Weyl data, and for \(k\) rounds the general bound
\[
F(\mathcal P_{\mathrm{rec}}(N,d,k))
\ge 1-2k\bigl(1-F(\mathcal P_{\mathrm{rec}}(N,d,1))\bigr)
\]
shows that the total error grows at most additively in the number of rounds. An important nuance is that better teleportation performance does **not** imply better recycling performance: in qubits, the optimized deterministic scheme can recycle slightly worse than the non-optimized one, despite being the better teleportation protocol [2105.14886].

For local Pauli noise on qubit resource pairs, the noisy deterministic PBT channel remains analytically tractable. The resulting channel is exactly a Pauli channel that factors as the finite-\(N\) PBT depolarizing map composed with an effective Pauli channel induced by the noisy resource, with fidelities
\[
f(\Lambda_{\vec p})=\frac12+\frac12 q_N q_{\vec p},
\qquad
F(\Lambda_{\vec p})=\frac14+\frac34 q_N q_{\vec p}.
\]
This factorization isolates architecture-induced error from resource-noise-induced error and underlies bounds for port-based entanglement teleportation [2309.01550].

Under pure dephasing, deterministic qubit PBT exhibits a different phenomenon. With noisy resource states but the original noiseless Ishizaka–Hiroshima measurement, the entanglement fidelity has an explicit closed form and converges, for large \(N\), to a dephasing-limited plateau determined by the real part of the coherence factor:
\[
F\to \frac{1+\Re\Gamma}{2}
\qquad (N\to\infty).
\]
More strikingly, when one replaces the original measurement by the pretty-good measurement adapted to the noisy ensemble, the adapted measurement performs worse than the noiseless one for essentially all finite \(N>2\). This is a concrete example where “noise adaptation” is not aligned with optimal operational performance [2602.16513].

## 6. Algorithms, applications, and related protocols

PBT is operationally important because Bob may process all ports before learning Alice’s classical message and later keep only the correct output wire. This is the basic asynchronous feature formalized in later resource-theoretic work: among communication models compatible with asynchronous quantum information processing, the strongest PBT-like model is already as powerful as arbitrary one-way teleportation for surpassing the classical teleportation threshold [2504.12945].

From an algorithmic perspective, efficient implementation was long missing. An initial breakthrough gave the first efficient quantum algorithm for deterministic PBT with the standard resource and pretty-good measurement, using twisted Schur–Weyl duality and the twisted Schur transform, with polynomial time in \(n\) and \(d\), \(O(n^{2d})\) space, and constant spectral-norm error [2310.01637]. Subsequent work provided explicit efficient circuits for probabilistic and deterministic PBT, both for EPR and optimized resource states, with two encodings of the Gelfand–Tsetlin basis. For constant local dimension and target error, the standard encoding achieves \(\widetilde O(n)\) time and \(O(n\log n)\) space, while the Yamanouchi encoding achieves \(\widetilde O(n^2)\) time and \(O(\log n)\) space [2312.03188].

The protocol is also a programmable simulator. For qubits, the Choi matrix of the channel simulated by PBT with a general resource state can be written explicitly in terms of that resource, so standard qubit PBT with variable program state becomes a universal simulator of qubit channels. Finite-port optimization of the resource improves concrete simulations, including amplitude damping, beyond the naive choice of taking \(N\) copies of the target channel’s Choi state [1912.10374].

Several nonstandard descendants of PBT clarify which parts of its complexity arise from correction-free decoding and which arise from universality over unknown inputs. Port-based telecloning replaces the single-port output by an \(M\)-subset of ports and asymptotically reaches the optimal universal \(1\to M\) cloning fidelity without any receiver correction unitaries [2501.16878]. Port-Based State Preparation, where Alice has a complete classical description of the target state, achieves error \(\left(1-\frac{1}{d}\right)^N\) with maximally entangled resources and this scaling is optimal for EPR resources; this strongly suggests that the polynomial scaling of ordinary PBT is tied to the unknown-input requirement rather than to port selection alone [2402.18356].

Taken together, these results place PBT at the intersection of teleportation theory, state discrimination, Schur–Weyl representation theory, programmable processing, and constrained quantum communication. Its defining operational move—replacing receiver-side correction by port selection—has generated a mathematically explicit theory with exact finite-\(N\) formulas, sharp asymptotics, nontrivial noise and recycling phenomena, and, more recently, efficient circuit constructions [1612.09260][1809.10751].

Source: https://www.emergentmind.com/topics/port-based-teleportation