🔢 mathematics

Membership in the cone of quantum states over time is NP-hard

This paper proves that determining membership in the convex cone of quantum states over time is NP-hard by demonstrating that the cone contains a linear copy of the bipartite separable cone, thereby establishing that no polynomial-time membership tests exist unless P=NP and that the cone lacks a finite semidefinite representation under specific dimensional constraints.

Original authors: Matt Hoogsteder-Riera

Published 2026-10-27

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Matt Hoogsteder-Riera

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quantum world, time does not always flow like a river moving in a single, predictable direction. Physicists have long sought a way to describe how a quantum system changes from one moment to the next, much like tracking a particle's position at two different times. For simple, classical systems, this is straightforward: you have a starting probability and a set of rules for how it changes, and you can map the entire history. But for quantum systems, where particles can exist in multiple states at once and influence each other in strange ways, this mapping becomes incredibly difficult. The mathematical objects used to describe these "states over time" are not always positive numbers; they can be negative or complex, behaving more like shadows than solid objects. This creates a vast, abstract landscape of possibilities. Scientists need to know which points in this landscape are physically real and which are impossible, a task that is crucial for understanding quantum communication, memory, and the very nature of cause and effect in the quantum realm.

A researcher has now shown that navigating this landscape is fundamentally harder than anyone previously realized. They proved that determining whether a specific quantum state over time is physically possible is a problem so complex that no computer could solve it quickly for all cases, unless a major breakthrough in computer science occurs (specifically, if P equals NP). This is not just a minor inconvenience; it is a mathematical wall. They demonstrated that the set of all possible quantum states over time contains a hidden structure that is identical to the set of "separable" states—those that do not exhibit the strange, interconnected behavior known as entanglement. Because checking for separability is already known to be one of the hardest problems in computer science, this discovery means that checking for valid quantum states over time is equally difficult. They did not just guess at this difficulty; they constructed a precise mathematical bridge that translates any question about separability into a question about these time-based states. If you could solve the time-based problem quickly, you could instantly solve the separability problem, which is widely believed to be impossible for standard computers.

They built this bridge by focusing on a specific setup where a quantum system evolves and then interacts with a copy of itself. They identified a special "witness," a mathematical tool that acts like a filter, separating the impossible states from the possible ones. When they applied this filter, they found that the states that passed through it perfectly matched the structure of separable states. This was a revelation because it meant the complex geometry of time-based quantum states was not just a new shape, but one that contained the entire difficulty of the separability problem within it. They showed that this relationship holds true even when the systems are small, provided they meet certain size requirements. They also proved that this set of states cannot be described by a finite list of simple rules that a computer could use to solve the problem efficiently via semidefinite programming. While the problem is technically decidable using real algebraic methods, they showed that no finite, efficient description exists for systems of a certain size, meaning we must rely on methods that are inherently slow and computationally expensive.

The implications of this work extend beyond just checking a list of possibilities. It suggests that the universe of quantum states over time is far more rugged and intricate than previously thought. They found that even small errors in measuring a state can make it impossible to tell if it is valid or not, a property known as instability. They calculated exactly how much a state can drift before it becomes indistinguishable from an impossible one, showing that this margin shrinks rapidly as the system gets larger. This instability is not a flaw in the theory but a fundamental feature of the geometry of quantum time. The work also rules out the possibility that these states can be neatly packaged into a finite set of equations that a computer could solve in a reasonable amount of time. This is a significant finding because many complex problems in physics are solved by finding such finite descriptions. Here, they proved that no such description exists for systems of a certain size, meaning we must rely on methods that are inherently slow and computationally expensive.

This discovery does not mean we cannot study quantum states over time, but it sets a clear boundary on what we can expect from our tools. It tells us that there is no shortcut to knowing if a quantum process is valid; we must pay the full computational cost. They did not find a way to bypass this difficulty, nor did they suggest that the problem is unsolvable in principle. Instead, they showed that the difficulty is real and unavoidable. The work stands as a definitive proof that the complexity of quantum time is deeply rooted in the mathematics of the universe. It leaves us with a clearer, albeit more challenging, picture of the quantum world: a place where the rules of cause and effect are not just hard to calculate, but are fundamentally resistant to the kind of quick, efficient solutions that make modern computing so powerful. The path forward will require new ways of thinking about complexity and perhaps new kinds of computers, but for now, the barrier is firmly established.

Technical Summary: Membership in the Cone of Quantum States over Time is NP-hard

Problem Statement
The paper addresses the computational complexity of characterizing the convex cone $\mathcal{Q}$ of "states over time" (stotes). A stote is defined as a Hermitian, generally non-positive operator on a bipartite system $H_A \otimes H_B$ constructed from an initial state $\rho$ and a quantum channel $E$ via the Jordan product prescription:
$Q = (\rho \otimes \mathbb{1}) \star J_E = \frac{1}{2}((\rho \otimes \mathbb{1})J_E + J_E(\rho \otimes \mathbb{1}))$
where $J_E$ is the Jamiołkowski operator of the channel $E$ . The set $\mathcal{Q}$ is the convex cone generated by all such operators. The central open problem, posed in [HCW26], is to characterize the membership properties of this cone and its dual $\mathcal{Q}^*$ . Specifically, the paper investigates whether determining if a given operator belongs to $\mathcal{Q}$ (exact membership) or is close to $\mathcal{Q}$ (weak membership) can be solved in polynomial time.

Methodology
The author employs a reduction from the quantum separability problem, which is known to be NP-hard. The core technical strategy involves constructing a facial embedding of the cone of separable states into the cone of states over time.

Facial Embedding Construction:
The author defines a specific output system $H_B = H_{A'} \otimes H_R$ , where $H_{A'}$ is a copy of the input space $H_A$ and $H_R$ is an auxiliary register. They construct a witness operator $W_0 = (d\mathbb{1}_{AA'} - S_{AA'}) \otimes \mathbb{1}_R$ , where $S_{AA'}$ is the swap operator.
They prove that $W_0$ belongs to the dual cone $\mathcal{Q}^*$ and that the exposed face of $\mathcal{Q}$ defined by the hyperplane orthogonal to $W_0$ (i.e., $\{Q \in \mathcal{Q} : \text{Tr}[W_0 Q] = 0\}$ ) is isomorphic to the cone of separable states $\text{SEP}^+(A:R)$ .
The Embedding Map:
An explicit, injective, rational linear map $L$ is defined such that for any Hermitian operator $\sigma$ on $A \otimes R$ :
$\sigma \in \text{SEP}^+(A:R) \iff L(\sigma) \in \mathcal{Q}(A:A'R)$
The map $L$ is constructed using the symmetrization map $\Gamma(X) = \frac{1}{2}\{X \otimes \mathbb{1}, S\}$ tensored with the identity on $R$ .
Stability and Robustness:
To extend the hardness result from exact membership to weak membership (the standard complexity notion for convex sets), the author derives a quantitative stability bound (Proposition 3.6). This bound relates the trace-norm distance of an operator $Y \in \mathcal{Q}$ from the exposed face to the value $\text{Tr}[W_0 Y]$ . This ensures that if an operator is far from the separable cone, its image under $L$ remains far from the stote cone, even after accounting for the "NO" instances in a promise problem.
Mixing Step:
To handle the boundary cases required for weak membership reductions, the author introduces a mixing step $T_\theta(X) = (1-\theta)X + \theta \mathbb{1}/N$ . This moves the image of the separable cone into the relative interior of the trace-one base of $\mathcal{Q}$ while preserving the separation for non-separable inputs.

Key Contributions and Results

NP-Hardness of Exact Membership:
Theorem 4.1 establishes that deciding whether a rational Hermitian operator belongs to $\mathcal{Q}$ is NP-hard. This is proven by the exact equivalence between separability and membership in the exposed face of $\mathcal{Q}$ .
NP-Hardness of Weak Membership and Optimization:
Theorem 4.2 proves that weak membership (determining if an operator is within a polynomially small distance of $\mathcal{Q}$ or far from it) is strongly NP-hard. Consequently, weak optimization over $\mathcal{Q}$ and its dual is also NP-hard. This implies that no polynomial-time membership tests exist for $\mathcal{Q}$ unless P = NP.
Non-Representability by Finite SDP:
Theorem 4.5 demonstrates that for dimensions $d, r \ge 2$ with $d+r > 5$ , the cone $\mathcal{Q}(A:A'R)$ admits no finite semidefinite representation (SDR). The proof relies on the fact that exposed faces and linear images of spectrahedral shadows are themselves spectrahedral shadows. Since the separable cone is known not to admit a finite SDR (by Fawzi's theorem [Faw21]), and $\mathcal{Q}$ contains a linear copy of it as a face, $\mathcal{Q}$ cannot be finitely SDR-representable.
Geometric Properties:
The paper establishes that $\mathcal{Q}$ is a closed, pointed, and spanning cone. It also provides a quantitative stability bound showing that the distance to the separable face scales with the square root of the witness pairing value ( $\sqrt{\eta}$ ), which is shown to be optimal.

Significance and Claims

The paper claims to resolve the complexity status of the membership problem for states over time, a structure central to quantum Bayesian inference, retrodiction, and quantum optimal transport.

Complexity Barrier: The primary significance is the demonstration that characterizing the cone of states over time is computationally intractable in the general case. This rules out the existence of efficient (polynomial-time) algorithms for membership testing or optimization over this cone, assuming P $\neq$ NP.
Limitations on SDP Hierarchies: The non-representability result implies that the cone cannot be exactly described by a finite number of semidefinite constraints. This suggests that while hierarchies of SDPs (like the Doherty-Parrilo-Spedalieri hierarchy for separability) may converge, they cannot provide an exact, finite-level characterization of the cone.
Distinction from Prior Work: The author clarifies that their hardness results do not contradict previous semidefinite tests for multi-time state compatibility (e.g., [GBB26]), as those tests address a different set allowing for initial system-environment correlations and memory, rather than the specific convex cone generated by state-channel pairs defined here.

The paper concludes by noting that while the hardness is established for specific output dimensions ( $B = A' \otimes R$ ), the complexity for arbitrary fixed dimensions or constant gaps remains an open question. It also highlights that the stability exponent derived is optimal, limiting the precision of potential approximation algorithms.

Technical Summary: Membership in the Cone of Quantum States over Time is NP-hard

More like this