Polytopic Quantum Resource Theories (PQRT)

• PQRTs are quantum resource theories where the set of free states forms a finite convex polytope generated by extremal states, unifying frameworks like coherence and magic.
• The framework employs a tensorial representation that disentangles linear structure from inner-product geometry, enabling detailed analysis of free-state polytopes.
• It establishes key classification results by distinguishing rigid CPTP-isomorphism from universal CP-isomorphism, and extends to compositional categorical structures.

Searching arXiv for the specified paper and closely related context. arxiv_search.query Searching for arXiv id (Naseri et al., 29 May 2026). <tool name="arxiv_search.query"> {"query":"(Naseri et al., 29 May 2026)","max_results":5} </tool> Searching arXiv now. {"query":"(Naseri et al., 29 May 2026)","max_results":5} Polytopic quantum resource theories (PQRTs) are quantum resource theories in which the set of free states is a convex polytope generated by a finite set of extremal free states. In the formulation of "Polytopic Quantum Resource Theories: Geometry and Structures" (Naseri et al., 29 May 2026), a PQRT is any resource theory where the free states can be expressed as a convex combination of a set of quantum states, referred to as extremal states. This class includes some of the most studied resource theories, such as coherence and magic. The framework introduces a tensorial representation that separates linear structure from Hilbert-space inner product structure, defines homomorphism and isomorphism for comparing theories at the level of both free states and allowed transformations, proves rigidity and universality results, studies linearly independent PQRTs under the name basis-non-convexity, and extends the formalism to a compositional symmetric monoidal setting (Naseri et al., 29 May 2026).

1. Definition of a polytopic quantum resource theory

Let $\mathcal{H}$ be a finite-dimensional Hilbert space of dimension $d$. Fix a finite set of vertex or extremal free states

$$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$

such that none of the $\sigma_i$ can be written as a convex combination of the others. The corresponding PQRT $\mathcal{R}$ is defined by its free states and its maximal free operations (Naseri et al., 29 May 2026).

The free-state set is

$F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$

so every $\rho_{\mathrm{free}}\in F_s$ admits

$$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$

The maximal free operations are

$$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$

By Lemma 1, it suffices to check $\Lambda(\sigma_i)\in F_s$ for $d$0.

This definition places the geometry of the free-state set at the center of the resource theory. In particular, the standard resource theory of coherence in a fixed basis $d$1 is the case $d$2 with $d$3, while the magic theory is the case where $d$4 consists of the pure stabilizer states. The paper’s conclusion further presents PQRTs as unifying all resource theories whose free state set is a finite convex polytope, including coherence, magic, and imaginarity (Naseri et al., 29 May 2026).

2. Tensorial representation and geometric content

A central construction separates the linear structure of states from the Hilbert-space inner product. One chooses an abstract vector space $d$5 of dimension $d$6, a basis $d$7 for $d$8, and a dual basis $d$9 for $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$0 satisfying $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$1 (Naseri et al., 29 May 2026).

Any density operator $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$2 on $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$3 is then written as an element of $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$4 by

$$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$5

where $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$6. In this abstract picture the set of all density operators is

$$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$7

For a PQRT with pure vertices $$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$8, the free-state polytope is

$$S_{\mathrm{ver}}^N=\{\sigma_i \mid \sigma_i\in D(\mathcal{H}),\ i=1\ldots N\}$$9

Within this representation, the key geometric data sit entirely in the abstract polytope $\sigma_i$0:

• Vertices: $\sigma_i$1.
• Facets: linear inequalities separating the convex hull.
• Dimension: $\sigma_i$2 as affine dimension.

Theorem III.1 states that any PQRT is uniquely, up to unitary, given by the triple

$\sigma_i$3

with free states generated by

$\sigma_i$4

This tensorial view exposes the geometry of the free-state polytope while distinguishing it from the choice of inner-product structure. A plausible implication is that questions about resource origin and theory comparison can be reformulated in terms of polytope structure before imposing Hilbert-space-specific identifications.

3. Homomorphism, isomorphism, and physical equivalence

The framework addresses the question of when two resource theories should be regarded as physically equivalent. Consider two PQRTs

$\sigma_i$5

A CPTP-homomorphism $\sigma_i$6 is a pair of maps (Naseri et al., 29 May 2026)

$\sigma_i$7

where $\sigma_i$8 is CPTP, such that:

• $\sigma_i$9,
• $\mathcal{R}$0 for all $\mathcal{R}$1,
• $\mathcal{R}$2.

If $\mathcal{R}$3 and $\mathcal{R}$4 are invertible with inverse maps satisfying the same conditions, the theories are CPTP-isomorphic.

A weaker notion is CP-homomorphism. In that case $\mathcal{R}$5 is allowed to be merely CP-norm-non-increasing, and the intertwining condition holds only up to a positive scalar $\mathcal{R}$6:

$\mathcal{R}$7

Invertibility yields CP-isomorphism.

The distinction between CPTP-isomorphism and CP-isomorphism is structurally significant. Deterministic equivalence is stricter than stochastic equivalence: the former preserves the theory at the level of CPTP dynamics, whereas the latter identifies theories up to normalization. This resolves a potential misconception that equality of polytope combinatorics alone suffices for full physical equivalence; in the PQRT setting, the answer depends on whether the comparison is made in the CPTP or CP sense.

4. Classification theorems

Two structural theorems organize the theory-comparison problem (Naseri et al., 29 May 2026). The first is a rigidity result:

If two PQRTs are CPTP-isomorphic then they must be geometrically equivalent, in the sense that their vertex sets are related by a unitary.

In particular, there is no nontrivial CPTP-isomorphism between inequivalent polytopes. The paper’s conclusion summarizes this as a unique deterministic equivalence only when the polytopes are unitarily equivalent.

The second theorem is a universality result:

Any two PQRTs with the same number $\mathcal{R}$8 of pure extremal vertices are CP-isomorphic.

Equivalently, all $\mathcal{R}$9-vertex polytopic theories are operationally equivalent up to stochastic normalization. Concretely, the isomorphism is constructed as

$F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$0

where $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$1 is the invertible linear map sending one dual basis to the other, together with

$F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$2

The proof sketch proceeds by showing that any two sets of $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$3 linearly independent pure vertices in $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$4 can be related by an invertible linear map on $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$5, after which the map is extended to density operators by normalization and $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$6 is defined by conjugation.

These two theorems establish a sharp dichotomy. On the one hand, CPTP-isomorphism is rigid and unitary-geometric. On the other hand, CP-isomorphism is universal for pure $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$7-vertex PQRTs. This suggests that the physically relevant notion of sameness depends on whether normalization-preserving implementations are required.

5. Linearly independent PQRTs and basis-non-convexity

A distinguished subclass arises when the extremal free states form a basis of the real space of Hermitian operators. In that case, the extremal free states $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$8 are linearly independent, and the free-state set $F_s := \operatorname{conv} S_{\mathrm{ver}}^N,$9 is a full-dimensional polytope of dimension $\rho_{\mathrm{free}}\in F_s$0 (Naseri et al., 29 May 2026). The complementary resource is termed basis-non-convexity.

Every state $\rho_{\mathrm{free}}\in F_s$1 then admits a unique expansion

$\rho_{\mathrm{free}}\in F_s$2

This unique affine decomposition makes it possible to define a negativity measure:

$\rho_{\mathrm{free}}\in F_s$3

It satisfies

$\rho_{\mathrm{free}}\in F_s$4

The paper shows that $\rho_{\mathrm{free}}\in F_s$5 is monotonic under free operations since free operations act by a column-stochastic map on the $\rho_{\mathrm{free}}\in F_s$6-vector.

A second monotone is geometric:

$\rho_{\mathrm{free}}\in F_s$7

which is likewise a free-operation monotone.

These constructions are specific to the linearly independent setting. Because the extremal free states form a basis, coefficient negativity directly detects departure from the free polytope. This gives the basis-non-convexity theory a structurally transparent relation between polytope geometry, affine coordinates, and monotone construction.

6. Compositional and categorical structure

The paper extends PQRTs beyond single systems by introducing a full compositional framework. For each finite-dimensional system $\rho_{\mathrm{free}}\in F_s$8 one chooses a set $\rho_{\mathrm{free}}\in F_s$9 and imposes the consistency condition

$$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$0

One then defines

$$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$1

and

$$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$2

Under these definitions, the collection of all systems and free CPTP maps forms a symmetric monoidal subcategory of the category of all CPTP maps, and identities, swaps, sequential compositions, and parallel compositions of free maps remain free (Naseri et al., 29 May 2026).

In this setting, a compositional CPTP-homomorphism $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$3 is given by:

• a function $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$4 sending each system $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$5 in $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$6 to a system $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$7 in $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$8;
• for each $$\rho_{\mathrm{free}}=\sum_{i=1}^N p_i \sigma_i,\qquad p_i\ge 0,\ \sum_i p_i=1.$$9, a CPTP map $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$0 such that $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$1;
• a monoidal functor $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$2 on free operations, preserving sequential and parallel composition, with intertwining

$$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$3

CPTP-isomorphism becomes isomorphism of monoidal subcategories. The paper explicitly notes that one can lift the definition of CPTP-homomorphism to a monoidal functor plus a monoidal natural transformation, cf. categorical resource theories à la Coecke, et al.

Two examples illustrate the framework. Coherence in the two qubit bases $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$4 and $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$5 are CPTP-isomorphic via the unitary $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$6. More generally, any $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$7-vertex PQRT, for example any choice of $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$8 pure states, is CP-isomorphic to the standard $$F_O := \{\Lambda\in \mathrm{CPTP}\mid \Lambda(\sigma)\in F_s\ \forall \sigma\in F_s\}.$$9-level coherence theory. These examples instantiate the broader distinction between deterministic unitary equivalence and stochastic universality established by the classification theorems.