Tensor Product Structure in Quantum Systems

• Tensor product structure (TPS) is a framework that organizes composite quantum systems into smaller subsystems, defining entanglement and locality.
• Different TPS choices can change entanglement properties as basis transformations and local unitaries redefine subsystem partitions.
• TPS concepts are applied in quantum information, many-body physics, quantum chemistry, and topological order, influencing simulation and error correction methods.

A tensor product structure (TPS) is a mathematical and physical framework that organizes the state space of composite quantum (or classical) systems as a product of smaller subsystems. In quantum theory, the choices of TPS fundamentally govern definitions of subsystems, locality, entanglement, and the formulation of tensor network states. The flexibility—and sometimes ambiguity—in the choice of TPS is at the core of quantum information theory, quantum many-body physics, quantum chemistry, and even the study of topological quantum order and emergent phenomena.

1. Formal Definitions and Basic Properties

For a finite-dimensional complex Hilbert space $\mathcal{H}$ of total dimension $n = n_1 n_2$, a bipartite TPS is a specification of two “subsystems” $A$ and $B$ with Hilbert spaces $\mathcal{H}_A \cong \mathbb{C}^{n_1}$ and $\mathcal{H}_B \cong \mathbb{C}^{n_2}$, given via an isomorphism:

$$\Phi: \mathcal{H} \to \mathcal{H}_A \otimes \mathcal{H}_B.$$

Two isomorphisms $\Phi$, $\Phi'$ are TPS-equivalent if $\Phi \Phi'^{-1} = U_1 \otimes U_2$ for local unitaries $n = n_1 n_2$0 on $n = n_1 n_2$1 and $n = n_1 n_2$2 on $n = n_1 n_2$3. Thus, a TPS is an equivalence class of such isomorphisms.

Explicitly, if one fixes an orthonormal basis for $n = n_1 n_2$4 labeled by double indices $n = n_1 n_2$5, where $n = n_1 n_2$6, $n = n_1 n_2$7, then $n = n_1 n_2$8 defines a standard TPS, but different basis choices or labelings induce different TPSs (Soulas, 26 Jun 2025).

2. TPS Choice, Basis Transformations, and Entanglement Structure

A central feature of the TPS framework is that the entanglement properties of a given state depend on the TPS chosen. Given a standard TPS $n = n_1 n_2$9, any other TPS can be obtained by a global unitary $A$0 acting on $A$1, where the image of the original product basis $A$2 under $A$3 is declared as the “new” product basis. That is:

$A$4

Consequently, $A$5 is identified with $A$6 via $A$7.

A state trajectory $A$8 is said to be disentangled by a TPS if, for all $A$9, there exist local states $B$0 such that:

$B$1

This definition underpins the analysis of entanglement generation as a function of TPS and has explicit consequences for the representation of quantum evolution (Soulas, 26 Jun 2025).

3. TPS in Tensor Network States: MPS, TTS, MERA, PEPS

Various families of tensor network states rely on an underlying TPS and its generalizations:

• Matrix Product States (MPS): For a chain of $B$2 sites, $B$3, an MPS is constructed from a set of rank-3 tensors, providing an efficient parametrization with entanglement bounded by the virtual bond dimension $B$4 (0910.1130).
• Tree Tensor States (TTS): The Hilbert space is recursively decomposed using isometries on a tree structure, favoring critical chains with logarithmic entanglement scaling.
• Multiscale Entanglement Renormalization Ansatz (MERA): Incorporates layers of disentanglers and isometries reflecting scale invariance and criticality.
• Projected Entangled Pair States (PEPS): A higher-dimensional generalization applicable to lattices, where each tensor respects the TPS and contract structure of the lattice geometry.

In all such constructions, the physical meaning and computational properties of the ansatz are a direct consequence of the chosen TPS (0910.1130).

4. TPS Distance, Operator Spreading, and Quantum Dynamics

A geometric measure of the difference between two TPSs under time evolution is captured by the TPS-distance $B$5:

$B$6

where $B$7 is the space of local operators (direct sum of subsystem algebras $B$8), and $B$9 is the Hilbert-Schmidt norm difference of their projectors. $\mathcal{H}_A \cong \mathbb{C}^{n_1}$0 quantifies how much a unitary delocalizes the algebra of local observables.

• For a symmetric bipartition, $\mathcal{H}_A \cong \mathbb{C}^{n_1}$1 coincides with the standard entangling power.
• $\mathcal{H}_A \cong \mathbb{C}^{n_1}$2 iff the unitary is local up to permutations; $\mathcal{H}_A \cong \mathbb{C}^{n_1}$3 if the unitary is “maximally delocalizing” (e.g., 2-unitaries).
• In Hamiltonian dynamics, the short-time growth is governed by the local scrambling rates.
• TPS-distance effectively distinguishes between chaotic, integrable, fragmented, and localized dynamics (Andreadakis et al., 2024).

5. Constraints on TPS and Disentanglement: Existence and No-Go Theorems

A key question is whether, for a given time-evolving state, there exists a TPS in which the state remains unentangled at all times. Explicit counterexamples and a general no-go result demonstrate:

• For certain evolutions (e.g., a C-NOT gate on two qubits), there exists a nonstandard TPS in which the entire trajectory is separable; an explicit unitary change of basis renders the otherwise maximally entangled Bell state a product state (Soulas, 26 Jun 2025).
• For generic state trajectories $\mathcal{H}_A \cong \mathbb{C}^{n_1}$4 where the set $\mathcal{H}_A \cong \mathbb{C}^{n_1}$5 is linearly independent, no fixed TPS (i.e., no single unitary change of basis) can render the trajectory separable for all $\mathcal{H}_A \cong \mathbb{C}^{n_1}$6 (Soulas, 26 Jun 2025).
• This result provides a rigorous separation between “special” evolutions that admit disentangling TPS and “generic” ones—highlighting the dynamical rigidity of entanglement in most interacting systems.

6. TPS, Quantum Mereology, and Locality

In quantum mereology—the study of part-whole relations in quantum systems—the TPS is essential for specifying what counts as a subsystem. The choice of TPS determines which subsystems are “local,” the structure of their algebras, and notions of causal influence.

• There is typically no dynamical or intrinsic principle that singles out a “preferred” TPS; for most interacting systems, no choice of TPS prevents entanglement growth during evolution.
• For Hamiltonians, this means that in general one cannot find a global TPS that turns an interacting Hamiltonian into a sum of local terms.

Applications include:

• Understanding quantum simulation and the emergence of locality in quantum gravity: TPS structure underpins what one means by “local Hamiltonian.”
• Reference-frame-dependent cryptography and error correction: alternative TPSs can “hide” or “expose” correlations and entanglement (Soulas, 26 Jun 2025).

7. Extensions: TPS in Quantum Chemistry, Hypergraph Theory, and Topological Phases

Quantum Chemistry & Correlation Methods:

The TPS framework is foundational in methods such as coupled-electron-pair approximations (CEPA) and tensor product selected configuration interaction (TPSCI). By partitioning orbitals into clusters and forming tensor products of their many-body eigenstates, TPS methods avoid singularities inherent to determinant-based expansions, enforce strict size-extensivity, and enable efficient, interpretable correlation treatments in both closed and open-shell systems (Abraham et al., 2022, Braunscheidel et al., 2024).

Hypergraph Structure:

In high-order data analysis, a TPS compatible with hypergraph structure defines operator products and Laplacians generalizing those in graph theory. This is critical for defining and optimizing algebraic connectivity, with implications for network robustness and spectral embeddings (Gu et al., 2023).

Topological Order and Gauge Symmetry:

For states with intrinsic topological order, such as the $\mathcal{H}_A \cong \mathbb{C}^{n_1}$7 toric code, the TPS is intricately tied to virtual gauge symmetry. Only variations of local tensors that preserve the virtual $\mathcal{H}_A \cong \mathbb{C}^{n_1}$8 symmetry maintain nonzero topological entanglement entropy; arbitrary deformations destroy topological order (Chen et al., 2010). Thus, the TPS must encode not only the subsystem decomposition but also the symmetry structure essential for topological protection.

---

In all these contexts, the tensor product structure is not merely a representation-theoretic convenience—it is a physically substantive choice that shapes the description, analysis, and control of complex quantum systems. The contemporary literature rigorously investigates both the flexibility and the constraints inherent to the TPS, revealing its central role in quantum information, condensed matter, quantum chemistry, and mathematical physics (Soulas, 26 Jun 2025, Andreadakis et al., 2024, 0910.1130, Abraham et al., 2022, Braunscheidel et al., 2024, Gu et al., 2023, Chen et al., 2010).