---
title: Quantum Texture-Based Resource Monotones
url: https://www.emergentmind.com/topics/texture-based-resource-monotone
type: topic
---

# Quantum Texture-Based Resource Monotones

Texture-based resource monotones are basis-dependent quantifiers of quantum-state “texture,” understood as the inhomogeneity of a state’s matrix-element distribution relative to a chosen computational or laboratory basis. In the original formulation, the unique textureless free state is the equal-superposition state, and the central monotone is the logarithmic “rugosity” obtained from the overlap with that free state. Subsequent work generalized this construction in two directions: from the fixed equal-superposition reference to an arbitrary pure reference state or a convex free set, and from the original grand-sum shortcut to fidelity-based and convex-roof formulations. These generalizations place texture within a broader class of fixed-point resource theories and connect it operationally to gate identification, especially the discrimination of CNOT-containing layers from single-qubit-only layers [2409.06482; 2602.22496].

## 1. Foundational definition and basis dependence

Fix a computational basis \(B=\{|i\rangle\}_{i=1}^D\). The original texture theory singles out the equal-superposition state
\[
|f_1\rangle = \frac{1}{\sqrt D}\sum_{i=1}^D |i\rangle,
\]
with free state \(f_1=|f_1\rangle\langle f_1|\). In the notation of the original texture work, the “grand sum” of matrix elements is
\[
\Sigma(\rho;B)=\sum_{i,j=1}^D \rho_{ij}=D\langle f_1|\rho|f_1\rangle,
\]
and the corresponding logarithmic monotone, called rugosity, is
\[
R(\rho;B)=-\ln\!\left(\frac{\Sigma(\rho;B)}{D}\right)=-\ln\langle f_1|\rho|f_1\rangle.
\]
For pure states \(|\phi\rangle\), this reduces to
\[
R(|\phi\rangle;B)=-\ln |\langle f_1|\phi\rangle|^2.
\]
The resource is basis dependent: changing the basis changes \(|f_1\rangle\), the grand sum, and hence the monotone itself [2409.06482].

The generalized formulation replaces \(|f_1\rangle\) by an arbitrary pure reference state \(|\psi\rangle\). One then defines
\[
\Sigma_\psi(\rho)=D\langle\psi|\rho|\psi\rangle,
\qquad
\mathcal R_\psi(\rho)=-\ln\!\left(\frac{\Sigma_\psi(\rho)}{D}\right)=-\ln\langle\psi|\rho|\psi\rangle.
\]
For pure inputs \(\rho=|\phi\rangle\langle\phi|\),
\[
\mathcal R_\psi(|\phi\rangle)=-\ln |\langle\psi|\phi\rangle|^2.
\]
This formulation preserves the overlap-based structure of the original rugosity while making explicit that the grand-sum shortcut is not essential [2602.22496].

The same body of work also emphasizes that the unique textureless state in the original theory is the only state whose density matrix has all entries equal in the chosen basis, and that \(R(\rho;B)=0\) if and only if \(\rho=f_1\). Conversely, \(R(\rho;B)\to\infty\) whenever \(\langle f_1|\rho|f_1\rangle\to 0\), as occurs for Fourier-basis states orthogonal to \(|f_1\rangle\) [2409.06482].

## 2. Resource-theoretic formulation

For a single reference free pure state \(|\psi\rangle\), the free completely positive trace-preserving maps \(\Lambda^\psi\) are those that fix \(\psi\):
\[
\Lambda^\psi(|\psi\rangle\langle\psi|)=|\psi\rangle\langle\psi|.
\]
Equivalently, every Kraus operator satisfies
\[
K_n^\psi|\psi\rangle=a_n^\psi|\psi\rangle,
\qquad
\sum_n |a_n^\psi|^2=1.
\]
In the decomposition \(\mathcal H=\mathcal H_\psi\oplus \mathcal H_\psi^\perp\), these Kraus operators have block form
\[
K_n^\psi=
\begin{bmatrix}
a_n^\psi & T_n^\psi\\
0 & L_n^\psi
\end{bmatrix},
\]
where \(T_n^\psi\) only couples states into \(\psi\), and \(L_n^\psi\) acts within \(\mathcal H_\psi^\perp\) [2602.22496].

This single-reference theory extends naturally to fixed-point resource theories. A fixed-point resource theory is specified by an orthogonal set of free pure states \(\{|\psi_i\rangle\}_{i=1}^m\) and by free CPTP maps \(\Lambda\) satisfying
\[
\Lambda(\sigma)=\sigma \quad \text{for all } \sigma\in \mathcal F_o=\operatorname{conv}\{|\psi_i\rangle\langle\psi_i|\}.
\]
In the \(\mathcal F_o\)-diagonal basis, Kraus operators take the block-triangular form
\[
K_n=
\begin{bmatrix}
\operatorname{diag}(\alpha_{1,n},\ldots,\alpha_{m,n}) & T_n\\
0 & L_n
\end{bmatrix},
\]
with \(\sum_n |\alpha_{i,n}|^2=1\) for each \(i\). These maps can destroy the resource by moving amplitude from \(\mathcal H_{\mathcal F_o}^\perp\) into \(\mathcal H_{\mathcal F_o}\) [2602.22496].

Several standard theories appear as fixed-point instances. In genuine coherence, the free states are diagonal in a chosen basis,
\[
\sigma_{\rm inc}=\sum_i p_i |i\rangle\langle i|,
\]
and free maps preserve all incoherent states, \(\Lambda_{\rm gi}(\sigma_{\rm inc})=\sigma_{\rm inc}\); the Kraus operators are fully diagonal, and dephasing is free. In purity, the free set is the maximally mixed state \(\tau=I/D\), and depolarizing maps \(\Lambda(\rho)=\tau\) are free and resource destroying. In athermality, the free set is the Gibbs state \(\tau_\beta\), and fixed-point maps preserve \(\tau_\beta\) [2602.22496].

## 3. Fidelity-based and convex-roof texture monotones

The most direct generalization from a single free state to a convex free set \(\mathcal F\) is the fidelity-based lower bound
\[
\underline{\mathcal R}_{\mathcal F}(\rho)
=
-\ln \max_{\sigma\in\mathcal F} F(\rho,\sigma),
\]
where the Uhlmann fidelity is
\[
F(\rho,\sigma)=\left(\operatorname{Tr}\sqrt{\sqrt\rho\,\sigma\,\sqrt\rho}\right)^2.
\]
For pure \(\rho=|\phi\rangle\langle\phi|\), this becomes
\[
\underline{\mathcal R}_{\mathcal F}(|\phi\rangle)
=
-\ln \max_{\sigma\in\mathcal F}\langle\phi|\sigma|\phi\rangle,
\]
so it coincides with the pure-state rugosity. Using the joint concavity of \(\sqrt F\) and the convexity and monotonicity of \(-\ln(\cdot)\), this quantity is convex in \(\rho\) and is a bona fide resource monotone [2602.22496].

A second extension is the convex-roof logarithmic texture measure. First define, for pure states,
\[
\mathcal R_{\mathcal F}(|\phi\rangle)
=
-\ln \max_{|\psi_k\rangle\in\mathcal F} |\langle\phi|\psi_k\rangle|^2.
\]
Then extend to mixed states by
\[
\mathcal R_{\mathcal F}(\rho)
=
\inf_{\rho=\sum_i p_i |\phi_i\rangle\langle\phi_i|}
\sum_i p_i\,\mathcal R_{\mathcal F}(|\phi_i\rangle).
\]
This construction preserves the logarithmic form and does not assume a grand-sum shortcut [2602.22496].

A distinct line of development introduces a two-parameter \(\alpha\)-\(z\) Rényi-based texture monotone. With \(f_1\) as the unique free state, the quantity
\[
\mathcal T^{\mathrm{GR}}_{\alpha,z}(\rho)
=
1-f_{\alpha,z}(f_1\Vert \rho)
=
1-\bigl(\langle f_1|\rho^{\frac{1-\alpha}{z}}|f_1\rangle\bigr)^z
\]
is a valid texture measure for
\[
\alpha\in(0,1),
\qquad
z\ge \max\{\alpha,1-\alpha\}.
\]
In that parameter regime it satisfies non-negativity, monotonicity under free operations, and convexity. The same work establishes unitary invariance under \(f_1\)-preserving unitaries and the tensor-product bounds
\[
\mathcal T^{\mathrm{GR}}_{\alpha,z}(\rho)+\mathcal T^{\mathrm{GR}}_{\alpha,z}(\delta)
\ge
\mathcal T^{\mathrm{GR}}_{\alpha,z}(\rho\otimes\delta)
\ge
\mathcal T^{\mathrm{GR}}_{\alpha,z}(\rho)\,\mathcal T^{\mathrm{GR}}_{\alpha,z}(\delta)
\]
[2604.07257].

These three constructions—rugosity, fidelity lower bound, and \(\alpha\)-\(z\) Rényi texture—share a common structural feature: each quantifies texture through a privileged overlap with a free state or free set. This suggests that overlap-based geometry, rather than the grand sum specifically, is the central organizing principle of the theory.

## 4. Monotonicity results and fixed-point phenomena

For fixed-point free sets \(\mathcal F_o\), define
\[
M_F(\rho)=\underline{\mathcal R}_{\mathcal F_o}(\rho)
=
-\ln \max_{\sigma\in\mathcal F_o} F(\rho,\sigma).
\]
If \(\Lambda\) is free and fixes every \(\sigma\in\mathcal F_o\), then fidelity data processing gives
\[
F(\Lambda(\rho),\sigma)=F(\Lambda(\rho),\Lambda(\sigma))\ge F(\rho,\sigma).
\]
Maximizing over \(\sigma\in\mathcal F_o\) and applying \(-\ln(\cdot)\) yields
\[
M_F(\Lambda(\rho))\le M_F(\rho),
\]
so the fidelity-based lower bound is weakly monotonic under free operations. Nonnegativity follows because \(\max_{\sigma\in\mathcal F_o}F(\rho,\sigma)\le 1\), with equality if and only if \(\rho\in\mathcal F_o\) [2602.22496].

By contrast, strong monotonicity can fail for the convex-roof logarithmic measure. In the one-reference case \(\mathcal F_o=\{|1\rangle\}\) in dimension \(D\), consider the pure state
\[
|\tau\rangle
=
\sqrt a\,|1\rangle
+
\sqrt{\frac{1-a}{D-1}}\sum_{k=2}^D |k\rangle,
\qquad
\frac12\le a\le 1,
\]
for which
\[
\mathcal R_{\mathcal F_o}(|\tau\rangle)=-\ln a.
\]
With diagonal Kraus operators
\[
K_1=\operatorname{diag}\!\left(\sqrt{\frac{1-a}{a(D-1)}},1,\ldots,1\right),
\qquad
K_2=\sqrt{I-K_1^\dagger K_1},
\]
the postselected outputs are \(|T_1\rangle\propto \sum_{k=1}^D |k\rangle\), having rugosity \(\ln D\), with probability
\[
p_1=\frac{D(1-a)}{D-1},
\]
and \(|T_2\rangle=|1\rangle\), having rugosity \(0\), with probability \(p_2=1-p_1\). Strong monotonicity would require
\[
-\ln a \ge p_1 \ln D,
\]
but this is violated whenever
\[
a\,D^{D(1-a)/(D-1)} > 1.
\]
An explicit family of violations is obtained for \(a=1-1/(2D)\), which yields violations for \(D=2,3,10,\ldots\) [2602.22496].

The same work notes that strong monotonicity is sensitive to the choice of free operations and basis, and that in the present fixed-point framework it fails generically for logarithmic convex-roof measures. Whether \(\mathcal R_{\mathcal F_o}\) is always weakly monotonic for all fixed-point theories remains an open question; numerical evidence suggests yes in the tested settings [2602.22496].

## 5. Operational role in gate identification

Texture-based monotones were first linked to gate identification through a protocol for distinguishing universal circuit layers that contain at least one CNOT from layers composed only of single-qubit gates. The protocol uses identically prepared Haar-random single-qubit inputs and texture measurements on the output qubits, without tomography or ancillary systems [2409.06482].

In the generalized formulation, one works in a laboratory basis that specifies both the reference state \(\psi\) and its Hadamard transform \(\psi_H=H\psi\). For a single layer acting on identically prepared Haar-random qubits, the average grand sum obeys
\[
\overline{\Sigma_\psi^s}=1
\]
for single-qubit-only gates. If a CNOT is present, and the two inputs are \(|\chi\rangle\otimes |\chi\rangle\), then in an unknown basis \(\{|c\rangle,|c'\rangle\}\) the reduced control and target states satisfy
\[
\overline{\Sigma_\psi(\rho^C)}
=
1+\frac23 \operatorname{Re}\langle\psi|c\rangle\langle c'|\psi\rangle,
\qquad
\overline{\Sigma_\psi(\rho^T)}
=
\frac23\bigl[1+|\langle\psi|c\rangle|^2\bigr],
\]
and similarly
\[
\overline{\Sigma_{\psi_H}(\rho^C)}
=
1+\frac23 \operatorname{Re}\langle\psi_H|c\rangle\langle c'|\psi_H\rangle,
\qquad
\overline{\Sigma_{\psi_H}(\rho^T)}
=
\frac23\bigl[1+|\langle\psi_H|c\rangle|^2\bigr].
\]
Any deviation from \(1\) in either control or target indicates the presence of a CNOT [2602.22496].

The generalized analysis shows that the protocol succeeds in “nearly all laboratory bases.” Failure occurs only if \(\psi\) lies on a measure-zero great circle on the Bloch sphere characterized by
\[
\psi_\pm=\frac{|c\rangle\pm i|c'\rangle}{\sqrt2}
\]
and
\[
U=e^{i\mu}R_y(\pi/4)R_z(\nu_2)
\quad \text{such that} \quad
HU|\psi_-\rangle = U|\psi_+\rangle,
\]
which forces all four averages to equal \(1\). Thus, in the continuous limit with \(\psi\) drawn arbitrarily, the success probability is unity, with measure-zero exceptions [2602.22496].

The original gate-identification paper states the same operational content in a more concrete two-setting form. In the chosen reference basis and its Fourier basis, wires not participating in any CNOT satisfy \(\Sigma_{\rm out}=1\) and \(\Sigma'_{\rm out}=1\), while CNOT-participating wires have at least one of these quantities deviating from \(1\) by at least \(1/6\), a basis-independent separation that supports complete layer characterization [2409.06482]. This operational role is significant because it identifies texture not merely as a static state property but as a directly measurable witness of entangling action in unknown circuit layers.

## 6. Relations to other resource theories, detection methods, and broader generalizations

With appropriate choices of free sets, the convex-roof texture formalism recovers known single-qubit measures. For imaginarity, taking the free set to be the real density matrices in a fixed basis gives, for pure states,
\[
\mathcal R_{\mathcal F}(|\phi\rangle)
=
-\ln \lambda_{\max}(\operatorname{Re}(|\phi\rangle\langle\phi|))
=
-\ln\!\left[\frac{1+|r_{\rm real}|}{2}\right],
\]
where \(r_{\rm real}^2=n_x^2+n_z^2\). For a mixed qubit with Bloch vector \(r=(r_x,r_y,r_z)\),
\[
\mathcal R_{\mathcal F}(\rho)
=
-\ln\!\left[\frac{1+\sqrt{1-r_y^2}}{2}\right].
\]
For coherence in a fixed computational basis, one has
\[
\mathcal R_{\mathcal F}(|\phi\rangle)
=
-\ln \max_i |\langle i|\phi\rangle|^2,
\]
and for a mixed qubit,
\[
\mathcal R_{\mathcal F}(\rho)
=
-\ln\!\left[\frac{1+\sqrt{1-r_\perp^2}}{2}\right],
\qquad
r_\perp^2=r_x^2+r_y^2.
\]
In the qubit case these coincide with standard geometric coherence forms [2602.22496].

A different texture-derived monotone uses basis optimization. If \(T_B(\rho)=1-\langle s_1|\rho|s_1\rangle\) is the basis-dependent texture, then over all orthonormal bases
\[
T^{\max}(\rho)=1-\lambda_d^\downarrow,
\qquad
T^{\min}(\rho)=1-\lambda_1^\downarrow.
\]
Their difference
\[
\Delta T(\rho)=T^{\max}(\rho)-T^{\min}(\rho)=\lambda_1^\downarrow-\lambda_d^\downarrow
\]
and the scaled quantity
\[
P(\rho)=d\,[T^{\max}(\rho)-T^{\min}(\rho)]
=
d(\lambda_1^\downarrow-\lambda_d^\downarrow)
\]
define a valid purity monotone under unital maps in any dimension. The same work also introduces non-local texture, which for bipartite pure states satisfies
\[
T_{\rm NL}^{A_1:A_2}(|\Psi\rangle)=1-\lambda_1,
\]
equal to the geometric measure of entanglement, and extends this viewpoint to coherence, non-stabilizerness, and multipartite genuine entanglement through convex-roof constructions [2507.13862].

Texture can also be detected by witnesses. A Hermitian operator \(W\) is a texture witness if \(\operatorname{Tr}(Wf_1)\ge 0\) and there exists at least one texture state \(\rho\) such that \(\operatorname{Tr}(W\rho)<0\). A universal construction is
\[
W=\Delta_T(A)-A,
\qquad
\Delta_T(X)=\operatorname{Tr}(Xf_1)f_1.
\]
Among the examples, the witness
\[
W_1=f_1-I
\]
detects all texture states because
\[
\operatorname{Tr}(W_1\rho)=\langle f_1|\rho|f_1\rangle-1=-\mathcal T_F(\rho)\le 0,
\]
with strict negativity for every \(\rho\neq f_1\). Additional witness families include
\[
W_\theta=\cos\theta\,I+\sin\theta\,(2f_1-I)
\]
and
\[
W^{jk}_\varphi
=
\frac{2\cos\varphi}{d}I
-
\bigl(e^{i\varphi}|j\rangle\langle k|
+
e^{-i\varphi}|k\rangle\langle j|\bigr),
\]
the latter detecting phase-sensitive off-diagonal structure and, for \(\varphi=\pi/2,3\pi/2\), the sign of the imaginary part of off-diagonal entries [2604.07257].

In a broader resource-theoretic setting, the abstract framework of universally-combinable resource theories shows how monotones arise by pullback through order-preserving maps. Within that framework, a texture-based monotone can be instantiated either by a set-based feature map with sup/inf root monotones, or by a tuple-based map combined with a contraction or information-content monotone. The same framework covers “distance to the free set” constructions, commuting-map constructions of the form \(M(r)=f(r,\Phi(r))\), and weight or robustness measures derived from the convex-alignment 3-contraction [1912.07085]. This suggests that fidelity-based texture monotones belong to a wider monotone-construction paradigm rather than constituting an isolated resource-specific device.

Several limitations remain explicit. Texture measures are basis dependent and sensitive to the choice of reference state or free set; convex-roof computations in higher dimensions remain challenging; strong monotonicity fails in general for the logarithmic convex-roof measure; and asymptotic continuity is not addressed in the \(\alpha\)-\(z\) Rényi construction. These limitations coexist with a notable strength of the framework: the same overlap-based formalism accommodates single-reference texture, fixed-point theories, coherence, imaginarity, purity, athermality, and operational gate-detection tasks within a common mathematical language [2602.22496; 2604.07257; 2507.13862].

Source: https://www.emergentmind.com/topics/texture-based-resource-monotone