Visualizing entanglement in multiqubit systems

Visualizing entanglement in multiqubit systems

Background and motivation

The paper addresses an established problem in quantum information education: how to show the structure of a multiqubit state, particularly its entanglement, without requiring readers to reconstruct every step through linear algebra. A Bloch sphere supports reasoning about a single qubit, but extending comparable geometric intuition to several qubits is difficult. The authors discuss geometric and topological representations, Majorana representations, variants of entangled Bloch spheres, Bloch hyperspheres, product-operator visualizations, and knot-based models. They also distinguish abstract graphical languages such as ZX, ZW, and ZH calculi, which support reasoning about gates and entanglement but presuppose familiarity with their semantics, from explicit state representations. DROPS visualizations offer another approach based on operators and generalized Wigner functions, with an emphasis on correlations and dynamics.

Dimensional circle notation, or DCN, combines two existing ideas: representing complex coefficients with circles and placing computational basis states along axes associated with individual qubits. Earlier circle notation presents basis states in a row, while cube and hypercube representations already give qubits a spatial organization. The authors' central contribution is to connect this dimensional organization to a visual separability criterion and use it to explain quantum processes. The problem of determining pure-state separability is already understood mathematically; the motivation here is to make that structure visible while following a state through an algorithm. The educational rationale draws on learning with multiple external representations, so the diagrams are intended to complement formulas and circuit diagrams.

Visual encoding and geometric operations

For a pure state

$$|\psi\rangle=\sum_{i=0}^{2^n-1}\alpha_i|i\rangle, \qquad \sum_i|\alpha_i|^2=1,$$

DCN retains one circle for each computational basis state. The radius of the filled inner disk represents $|\alpha_i|$, making its area proportional to the measurement probability $|\alpha_i|^2$. A radial line represents the coefficient's phase through its angle from the vertical. An empty circle therefore denotes a zero coefficient, rather than an absent basis state. The paper numbers qubits from the rightmost entry in a ket, so qubit 1 is the least significant bit. Where ordinary circle notation lines up all $2^n$ circles, DCN arranges two qubits as a square and three as a cube, with neighboring vertices along a qubit's axis differing only in that qubit's bit value. Figure 1 introduces the radius and phase encoding, Figure 2 shows the conventional linear layout, and Figure 3 demonstrates how two single-qubit factors form a two-dimensional product state.

Figure 3 from Bley et al., Physical Review Research 6, 023077 (2024), DOI: 10.1103/PhysRevResearch.6.023077, reproduced under CC BY 4.0. The green line marks a common coefficient-ratio relation across both rows; the arrows distinguish magnitude scaling from phase rotation.

The spatial layout also gives gates a consistent interpretation. A single-qubit operation acts on every pair of coefficients connected along the corresponding axis. For example, an $X$ gate exchanges the paired coefficients, a $Z$ gate changes the phase of coefficients on the qubit's 1 side by $\pi$, and a Hadamard gate mixes the two coefficients in each pair. Figure 6 contrasts these structured exchanges with the same operations on a linear circle display. A CNOT exchanges coefficients along the target axis only where the control bit equals 1, while a SWAP exchanges the roles of two axes. Figure 7 shows how three CNOT operations produce that SWAP, even though the intermediate states can have different entanglement from the initial and final states. These are explanatory state diagrams accompanied by circuit diagrams, rather than screenshots of an evaluated application interface.

A computational-basis measurement selects the subset of circles compatible with its outcome, empties the incompatible circles, and renormalizes the remaining coefficients. The outcome probability is obtained by summing the relevant inner-disk areas. Figure 5 illustrates a two-qubit example with outcome probabilities $3/4$ and $1/4$, where the surviving state of the unmeasured qubit depends on the result. The spatial arrangement makes the subsets to be retained or discarded explicit.

Separability as a common coefficient ratio

For two qubits, the familiar pure-state separability condition is

$$\alpha_{00}\alpha_{11}=\alpha_{01}\alpha_{10}.$$

When the relevant denominators are nonzero, this condition says that the complex ratio across one row of the square must match the ratio across the other row. Both the radius-scaling factor and the phase difference must agree. The paper calls this a symmetry condition, but the two sides need not be identical mirror images: the correspondence can include a common magnitude scaling and phase rotation. Green lines indicate that the ratio condition holds, while red lines indicate that it fails. Figure 4 contrasts failure caused by unequal magnitude ratios with failure caused by unequal phase differences. The authors relate these differences to the two-qubit concurrence, $C=2|\alpha_{00}\alpha_{11}-\alpha_{01}\alpha_{10}|$, providing a mathematical connection between the visual discrepancy and entanglement. The magnitude and phase explanations refer to the displayed basis representation.

For three qubits, the same reasoning compares corresponding coefficients on opposite faces of the cube. A common ratio across all corresponding pairs shows that the qubit perpendicular to those faces factors from the remaining two-qubit subsystem. Figure 8 uses a green plane to show such a factorization and a red plane to show that the other qubits remain entangled. Thus, one valid plane can identify partial separability; satisfying the corresponding conditions for two distinct qubits implies full separability of the pure three-qubit state. This is a concrete way to distinguish a product of one qubit and an entangled pair from a product of three independent qubits.

The paper generalizes the argument to a chosen bipartition of a pure state in an $N=PQ$ dimensional space. Writing its coefficients as a $P\times Q$ array, $M_{kr}=c_{kQ+r}$, the state factors across that partition when the coefficient array is an outer product. The equivalent cross-product relations are

$$M_{k_0r_0}M_{kr}=M_{k_0r}M_{kr_0}.$$

Using a nonzero reference entry and checking all required entries avoids dividing by zero and exposes the common ratios that DCN depicts. Theorem 1 and Appendix B provide the derivation and alternative formulations, including the treatment of leading zero coefficients. This zero-coefficient issue is substantive: Figure 10 shows a state for which a selected equality holds even though the proposed partition is not separable, because another coefficient violates the required zero pattern. Figure 15 extends the reasoning to a four-qubit state that factors into two two-qubit subsystems, although no individual qubit factors from the whole system. Consequently, checking only single-qubit symmetry planes does not characterize every possible bipartition in larger systems. The construction uses a product basis, here the computational basis, and concerns pure states rather than a general solution for mixed-state separability.

Algorithm examples and modular layouts

Quantum teleportation provides the main three-qubit example. Figure 9 begins with an input qubit that is independent of a shared Bell pair, applies Alice's CNOT and Hadamard operations, and shows how the coefficient structure changes across the cube. Figure 14 then separates the four possible measurement branches, each with probability $1/4$ in the illustrated protocol, and identifies the conditional $X$ and $Z$ corrections Bob needs to recover the input state. The geometric account therefore includes both entangling operations and the measurement outcomes and classical communication required for teleportation. Appendix C also depicts a Hadamard-basis identity that reverses the control and target roles of a CNOT and gives a worked Deutsch-algorithm example. Its displayed states retain the separability relation, illustrating that the presence of a CNOT in a circuit does not itself imply that the particular input becomes entangled. The paper cites an earlier classical optical realization of that algorithm; it does not report such an experiment itself.

For four and five qubits, the authors allow the layout to change with the explanatory task. Four-qubit states can be shown as two connected cubes, a projection of a hypercube. Their modular DCN combines linear circle notation with selected spatial axes, assigning individual axes to qubits whose roles need emphasis while grouping other qubits together. Figure 16 visualizes an existing four-qubit error-detection procedure, including a Hadamard-error example whose two syndrome qubits are anticorrelated. Figures 17–19 illustrate a bit-flip correction procedure with three data qubits and two additional syndrome qubits, assuming at most one bit flip among the data qubits. The sequence first shows three-qubit encoding and alternative error cases, then flattens those eight basis states into one axis of a $2\times2\times8$ display, and finally rearranges the state into four cubes indexed by the syndrome bits. The colors distinguish the no-error and different single-error cases; they do not encode an additional state amplitude or an empirical frequency. For those illustrated error cases, the paper uses the layout to distinguish correlations between the error and its syndrome from entanglement within the encoded data subsystem. This example should be understood within its stated bit-flip model, rather than as a demonstration of arbitrary-error correction by the five-qubit perfect code.

Contributions, evidence, and limitations

The contribution is a mathematically supported interpretation of dimensional state diagrams, together with detailed examples of gates, measurements, separability across different partitions, and small quantum algorithms. The paper supports its claims with derivations, proofs, and worked visual examples, including direct diagrammatic comparisons with ordinary circle notation. It reports no controlled learner study, measured improvement in comprehension, task-performance comparison, or usability evaluation of DCN. Claims that the notation can make quantum concepts more intuitive and accessible are therefore proposed educational benefits. At publication, the authors reported ongoing development of an interactive web tool and linked repositories and a working beta version. The article does not establish a tested interaction workflow or measure the beta's effectiveness.

Explicitly showing $2^n$ basis states imposes an exponential visual cost. The authors expect systems larger than roughly six or seven qubits to be difficult to display, and acknowledge that more complicated bipartitions become harder to recognize even when the mathematical procedure remains valid. Visual inspection also cannot reliably establish exact equality of numerical amplitudes and phases, so the diagrams do not replace the underlying algebra. Displays without symbolic variables show particular examples, from which readers must abstract a general principle. These limits distinguish an exact separability theorem from the precision with which a reader can judge a rendered image. Future directions include further development of the interactive tool and extension to qudits, for which the authors suggest applying the same ratio characterization. The modular layouts provide flexibility for explaining selected structures, while retaining the basic limitation that the full pure-state coefficient set still grows exponentially.