Teleportation-Based Quantum State Tomography

• Teleportation-based quantum state tomography is a protocol that uses teleportation primitives to convert high-dimensional quantum states into measurable single-qubit outputs.
• It employs Bell and generalized Bell measurements with fixed input states to reconstruct either complete density matrices or selective matrix elements in qubit and qutrit systems.
• While the method redistributes measurement complexity and can provide exponential advantages for sparse states, full-state reconstruction still scales exponentially with system size.

Teleportation-based quantum state tomography (QST) denotes a class of protocols in which teleportation primitives are used to reconstruct quantum states by transferring state information into a form that is easier to measure. In one formulation, Bell measurements and a small fixed set of known single-qubit inputs suffice to completely reconstruct arbitrary two-, three-, and $n$-qubit density matrices (Rigolin, 23 Nov 2025). In another, a chosen $2\times2$ logical block of an $N$-qubit density matrix is nondestructively teleported onto a single physical prober qubit for direct readout of a selected matrix element or wave-function component (Chen et al., 2021). The same general program has been extended from qubits to spin-1 qutrits, where generalized Bell measurements and single-qutrit preparations yield complete reconstruction of arbitrary two-qutrit states (Rigolin, 6 Jul 2026).

1. Conceptual structure of teleportation-based QST

The central idea is to use teleportation not as a communication primitive for transmitting an unknown state through a known channel, but as a measurement primitive that converts inaccessible or high-dimensional information into experimentally manageable output states. In the spin-1 formulation, the protocol is explicitly described as one that reverses teleportation: instead of sending one unknown state through a known channel, one sends known states through the unknown channel and reads out the results by Bell measurements (Rigolin, 6 Jul 2026). In the direct-measurement formulation, the target object is not necessarily the whole density matrix at once, but a logical two-dimensional subspace spanned by computational-basis states $|m\rangle$ and $|n\rangle$, which is then teleported to a single prober qubit (Chen et al., 2021). In Rigolin-style full tomography, the unknown shared state itself acts as the teleportation channel, and repeated Bell measurements with known auxiliary inputs generate a linear system whose inversion yields the full density matrix (Rigolin, 23 Nov 2025).

These constructions differ in resource layout, but they share a common operational pattern. Alice prepares known inputs, performs Bell or generalized Bell measurements jointly on the input and part of the unknown system or an entangled resource, communicates the classical outcomes, and Bob reconstructs either a single output qubit, a single output qutrit, or a selected matrix element. This suggests a family rather than a single canonical circuit: the common element is teleportation-assisted extraction of density-matrix data, whereas the exact teleportation channel, readout system, and scaling depend on the target task.

A persistent conceptual distinction is between complete tomography and selective access. The qubit and qutrit Rigolin-style schemes are full reconstruction protocols. The logical-qubit method is selective by construction: it isolates the block

$$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$

teleports that block to a prober qubit, and measures only the chosen entry $\rho_{mn}$ or a sparse set of entries.

2. Rigolin-style complete reconstruction for qubits

For arbitrary two-qubit tomography, Alice prepares an auxiliary qubit $A$ in one of four known pure states spanning the single-qubit operator space,

$$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$

and performs a Bell-basis measurement on $(A,1)$. The Bell projectors are

$2\times2$0

with

$2\times2$1

Conditioned on outcome $2\times2$2, Bob’s qubit is

$2\times2$3

where $2\times2$4 and

$2\times2$5

Because Alice’s input is known, Bob need not apply the usual Pauli correction; the conditioned output itself is the tomographic data (Rigolin, 23 Nov 2025).

Writing the unknown two-qubit density matrix in the computational basis as

$2\times2$6

and Bob’s post-teleportation state as

$2\times2$7

the protocol produces explicit linear relations between the measured $2\times2$8-coefficients and the unknown matrix elements $2\times2$9. For a fixed Bell outcome such as $N$0, the relations include

$N$1

$N$2

$N$3

Choosing the four inputs $N$4 yields $N$5 equations, and together with normalization $N$6 one can solve uniquely for the $N$7 independent parameters of a general two-qubit state (Rigolin, 23 Nov 2025).

The same construction generalizes to $N$8 qubits. Alice prepares $N$9 auxiliary qubits in one of the $|m\rangle$0 product states drawn from $|m\rangle$1, performs $|m\rangle$2 independent Bell measurements on pairs $|m\rangle$3, and Bob performs single-qubit tomography on qubit $|m\rangle$4. Each run yields $|m\rangle$5 real parameters from one single-qubit output, so $|m\rangle$6 distinct inputs produce a linear system of size $|m\rangle$7, whose inversion reconstructs all $|m\rangle$8 real parameters of $|m\rangle$9 (Rigolin, 23 Nov 2025).

The significance of this qubit protocol is architectural rather than asymptotic. It replaces multi-qubit joint measurements on Bob’s side with Bell measurements on Alice’s side plus single-qubit tomography on Bob’s side. It is therefore most naturally viewed as a redistribution of tomographic difficulty, not a generic escape from exponential state-space growth.

3. Direct access to individual matrix elements via logical-qubit teleportation

A distinct teleportation-based approach targets a chosen matrix element rather than the full density matrix. Given computational-basis labels $|n\rangle$0 and $|n\rangle$1 of an unknown $|n\rangle$2-qubit density matrix $|n\rangle$3, one defines the logical two-dimensional subspace spanned by $|n\rangle$4 and $|n\rangle$5. Restricted to that subspace,

$|n\rangle$6

The aim is to nondestructively isolate this block and teleport it onto a single physical prober qubit $|n\rangle$7 (Chen et al., 2021).

The protocol uses a GHZ channel whose size depends on the Hamming distance $|n\rangle$8 between $|n\rangle$9 and $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$0. One prepares $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$1 qubits, consisting of ancillas $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$2 and the prober $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$3, in

$$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$4

For each position $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$5, let $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$6. If $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$7, system qubit $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$8 is measured immediately in the $$\rho^{(\mathrm{log})} = \begin{bmatrix} \rho_{mm} & \rho_{mn}\ \rho_{nm} & \rho_{nn} \end{bmatrix},$$9 basis. If $\rho_{mn}$0, a Bell measurement is performed on system qubit $\rho_{mn}$1 and an ancilla qubit $\rho_{mn}$2, implemented by a CNOT from system $\rho_{mn}$3 to $\rho_{mn}$4, a Hadamard on system $\rho_{mn}$5, and then $\rho_{mn}$6-basis measurements on both qubits. The total measurement record determines a Pauli correction

$\rho_{mn}$7

which is applied on the prober or tracked in post-processing (Chen et al., 2021).

After correction, the prober carries the normalized logical qubit

$\rho_{mn}$8

Single-qubit observables on $\rho_{mn}$9 then recover the selected off-diagonal entry: $A$0 so that

$A$1

The protocol therefore converts one targeted density-matrix element into a single-qubit readout problem (Chen et al., 2021).

This method yields a different scaling statement from full tomography. Standard tomography on $A$2 qubits requires at least $A$3 distinct measurement settings, whereas the direct-measurement protocol first measures all diagonal populations with a single $A$4 readout and then uses one teleportation circuit for each pair $A$5 of interest. If only $A$6 computational-basis populations exceed a threshold, there are at most $A$7 relevant off-diagonal pairs, so the number of settings is $A$8. The claimed exponential advantage is therefore explicitly tied to sparse states rather than arbitrary dense states (Chen et al., 2021).

4. Spin-1 and qutrit generalization

The spin-1 extension shows that the original qubit protocol can be generalized to qutrits. The required resources are a maximally entangled qutrit pair and the ability for Alice to prepare a few different single-qutrit states to be teleported to Bob (Rigolin, 6 Jul 2026). A canonical entangled resource is

$A$9

together with the nine orthonormal Bell-qutrit projectors

$$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$0

One convenient orthonormal maximally entangled basis is

$$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$1

with the remaining states obtained analogously from the $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$2 and $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$3 patterns with phases $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$4 and $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$5 (Rigolin, 6 Jul 2026).

Alice’s preparation set is described as the computational basis $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$6, the real superpositions $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$7, the phase superpositions $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$8, and similarly for the pairs $$|0\rangle,\quad |1\rangle,\quad |+\rangle\equiv \frac{|0\rangle+|1\rangle}{\sqrt2},\quad |R\rangle\equiv \frac{|0\rangle+i|1\rangle}{\sqrt2},$$9 and $(A,1)$0. The description states that these nine choices yield $(A,1)$1 real equations, sufficient to fix all $(A,1)$2 real parameters of a general two-qutrit density matrix (Rigolin, 6 Jul 2026).

For one run of the two-qutrit protocol, Alice chooses a known input

$(A,1)$3

and the initial state is

$(A,1)$4

She performs a Bell measurement on $(A,1)$5. If outcome $(A,1)$6 occurs, the probability is

$(A,1)$7

and Bob’s raw teleported qutrit state is

$(A,1)$8

By repeating the protocol for each input state and Bell outcome, Bob estimates the matrix elements $(A,1)$9 experimentally (Rigolin, 6 Jul 2026).

The reconstruction is linear. Defining unnormalized elements $2\times2$00, one obtains explicit linear combinations of the unknown two-qutrit matrix elements $2\times2$01. For outcome $2\times2$02, one example is

$2\times2$03

Teleporting $2\times2$04 fixes the three diagonal $2\times2$05 blocks of $2\times2$06, while teleporting the six superposition pairs generates enough independent equations to solve for all $2\times2$07 real parameters. An explicit example is

$2\times2$08

with analogous formulas for the remaining $2\times2$09 parameters (Rigolin, 6 Jul 2026).

The qutrit result is important because it shows that the teleportation-based strategy is not restricted to spin-$2\times2$10 systems. The same pattern is stated to extend to higher local dimension $2\times2$11: one uses $2\times2$12 maximally entangled basis states and $2\times2$13 input states from each local $2\times2$14-dimensional Hilbert space (Rigolin, 6 Jul 2026).

5. Resource requirements, scaling laws, and comparison with standard tomography

The principal resource statements differ across the three main formulations.

Variant	Core resources	Stated scaling
Rigolin-style full qubit QST	Four single-qubit input states, Bell measurements, Bob single-qubit tomography, no extra entangled pairs beyond the shared unknown state	$2\times2$15 runs, total settings $2\times2$16
Logical-qubit direct measurement	GHZ resource, Bell measurements on differing bit positions, prober-qubit $2\times2$17 readout	One $2\times2$18 setting for populations, then $2\times2$19 circuits for sparse states
Spin-1 qutrit QST	Maximally entangled qutrit pairs, Bell-qutrit measurements, fixed single-qutrit preparations	$2\times2$20 teleportation settings in the worst case

In the qubit full-reconstruction protocol, the unknown state itself serves as the teleportation channel, Alice sends $2\times2$21 classical bits per run corresponding to the Bell outcomes, and Bob needs only single-qubit tomography on the remaining qubit. Standard $2\times2$22-qubit tomography requires $2\times2$23 measurement bases on $2\times2$24 qubits, whereas teleportation-based QST requires $2\times2$25 runs and total settings $2\times2$26 (Rigolin, 23 Nov 2025). For large $2\times2$27, $2\times2$28 grows faster than $2\times2$29, so this formulation trades joint multi-qubit measurements for a larger number of single-qubit measurements rather than improving worst-case asymptotics.

The sparse direct-measurement protocol yields a different comparison. After a single $2\times2$30 population readout, each pair $2\times2$31 of interest requires exactly one teleportation circuit $2\times2$32, and the number of settings becomes $2\times2$33 when only $2\times2$34 basis populations are significant (Chen et al., 2021). This is the precise sense in which an exponential advantage is claimed: it applies to sparse multi-particle states, not to arbitrary full-density-matrix reconstruction.

In the qutrit extension, the worst-case statement matches standard tomographic dimensionality. The protocol is described as needing $2\times2$35 teleportation settings to reconstruct an arbitrary $2\times2$36-qutrit state, just as standard tomography requires measuring all $2\times2$37 generalized Pauli correlators (Rigolin, 6 Jul 2026). The qutrit generalization is therefore structurally significant—it replaces qubit Bell states and four inputs by qutrit Bell states and nine inputs—but it does not remove the worst-case exponential growth.

6. Experimental constraints, error models, and common points of confusion

All variants are sensitive to Bell-measurement fidelity, input-state preparation error, and finite-sample noise. In the qubit full-reconstruction protocol, imperfect Bell measurements introduce systematic biases in the reconstructed state, deviations of $2\times2$38 from their ideal Bloch vectors degrade the linear inversion, and finite statistics affect both the Bell-outcome probabilities and the single-qubit tomography data. Maximum-likelihood reconstruction may be applied to enforce positivity of the final density matrix (Rigolin, 23 Nov 2025).

The qutrit construction explicitly assumes perfect maximally entangled qutrit pairs and ideal projective Bell measurements. The accompanying practical note states that in reality one must calibrate for nonideal visibilities and finite statistics (Rigolin, 6 Jul 2026). Because the inversion is linear, systematic calibration errors propagate directly into the reconstructed $2\times2$39-parameter density matrix.

The direct-measurement scheme has been implemented experimentally for photonic mixed states. Three pairs of EPR photons were generated by three Type-II SPDC processes in $2\times2$40-barium-borate crystals pumped by a $2\times2$41, $2\times2$42, $2\times2$43 UV laser, with each pair nominally in $2\times2$44 with fidelity $2\times2$45. Bell measurements were realized by overlapping photons on a PBS so that Hong–Ou–Mandel interference and diagonal polarizers implemented a projection onto $2\times2$46, and the prober photon was measured in the $2\times2$47 or $2\times2$48 basis with QWP+HWP+PBS cascades and single-photon Si avalanche detectors. The reported teleportation fidelities were $2\times2$49–$2\times2$50 for the various logical-qubit classes (Chen et al., 2021).

A noisy GHZ resource of the form

$2\times2$51

attenuates the off-diagonal readout according to

$2\times2$52

so an independently calibrated $2\times2$53 permits renormalization of the measured quantity (Chen et al., 2021). The linear-optics implementation also distinguishes only one Bell state, $2\times2$54, so full four-outcome Bell analysis with feed-forward would improve heralding probability and reduce post-selection loss.

One common point of confusion is whether teleportation-based QST generically overcomes the curse of dimensionality. The literature does not support that claim in full generality. For arbitrary $2\times2$55-qubit or $2\times2$56-qutrit states, the complete reconstruction protocols remain exponential in system size (Rigolin, 23 Nov 2025). The sparse-state logical-qubit protocol is the exceptional case: its advantage is conditional on the existence of only $2\times2$57 significant computational-basis populations, in which case the number of relevant off-diagonal pairs is at most $2\times2$58 (Chen et al., 2021). Another recurring misconception is that these schemes eliminate entanglement resources altogether. That statement is true for the Rigolin-style qubit protocol, where no extra entangled pairs are needed beyond the shared unknown state, but it is not true for the qutrit extension or the logical-qubit direct-measurement protocol, both of which require explicit entangled resource states (Rigolin, 23 Nov 2025).