---
title: Zero-Uncertainty States with Quantum Memory
url: https://www.emergentmind.com/topics/zero-uncertainty-states-with-quantum-memory
type: topic
---

# Zero-Uncertainty States with Quantum Memory

Searching arXiv for the primary paper and closely related foundational work.
{"query":"2603.23036 Zero-Uncertainty States Relative to Observable Algebras 2003.02103 Zero Uncertainty States in the Presence of Quantum Memory Berta uncertainty quantum memory", "max_results": 10}
I found the following relevant arXiv papers:
- "Zero-Uncertainty States Relative to Observable Algebras" [2603.23036]
- "Zero Uncertainty States in the Presence of Quantum Memory" [2003.02103]
- "Experimental investigation of the uncertainty principle in the presence of quantum memory" [1012.0332]
- "Position-Momentum Uncertainty Relations in the Presence of Quantum Memory" [1308.4527]
- "Quantum discord and classical correlation can tighten the uncertainty principle in the presence of quantum memory" [1204.3803]


Zero-uncertainty states with quantum memory are bipartite states \( \rho_{AB} \) for which measurement outcomes on subsystem \(A\) become perfectly predictable to an observer holding subsystem \(B\). In the finite-dimensional setting, this perfect predictability is equivalent to vanishing conditional entropy for the measured observable and, operationally, to perfect distinguishability of the corresponding conditional operators on the memory system. The operator-algebraic treatment developed in "Zero-Uncertainty States Relative to Observable Algebras" extends the earlier nondegenerate theory to degenerate projective-valued measurements (PVMs), makes observable subalgebras explicit, and isolates the precise mechanisms by which zero uncertainty either rigidly enforces purity and maximal entanglement or fails to do so [2603.23036][2003.02103].

## 1. Definition and operational content

For a finite-dimensional Hilbert space \(H_A\), a PVM on \(H_A\) is a finite family of mutually orthogonal projections \(K=\{P_\alpha\}\subset B(H_A)\) with \(\sum_\alpha P_\alpha=I_A\). Degeneracy means that the projections \(P_\alpha\) can have rank \(>1\), so the measurement is coarse-grained onto subspaces rather than rank-one eigenspaces. For a bipartite state \(\rho\in B(H_A\otimes H_B)\), the conditional operators on Bob’s space are
\[
Z_\alpha:=\operatorname{Tr}_A[(P_\alpha\otimes I_B)\rho]\in B(H_B).
\]
The state \(\rho\) is a zero-uncertainty state for \(K\) if the family \(\{Z_\alpha\}_\alpha\) is perfectly distinguishable, which in finite dimensions is equivalent to orthogonality of supports,
\[
\operatorname{supp}(Z_\alpha)\perp \operatorname{supp}(Z_\beta)\quad \text{for all }\alpha\neq\beta,
\]
or, equivalently,
\[
Z_\alpha Z_\beta=0 \quad \text{for all }\alpha\neq\beta,\qquad \sum_\alpha Z_\alpha=\rho_B:=\operatorname{Tr}_A\rho.
\]
For a family \(\mathcal K=\{K_i\}\) of PVMs, \(\rho\) is a common ZUS if it is a ZUS for every \(K_i\) [2603.23036].

The information-theoretic interpretation is expressed through the post-measurement classical-quantum state
\[
\rho_{XB}=\sum_x |x\rangle\langle x|\otimes Z_x,
\]
with \(Z_x=\operatorname{Tr}_A[(P_x\otimes I)\rho]\). The classical conditional entropy \(H(X|B)\) vanishes exactly when the ensemble \(\{Z_x\}_x\) is perfectly distinguishable, so zero uncertainty for \(X\) relative to memory \(B\) means \(H(X|B)=0\). In the earlier nondegenerate formulation, a ZUS with respect to a set of observables \( \{O_k\} \) is equivalently a state for which, for every \(k\), there exists a POVM on \(B\) allowing deterministic prediction of Alice’s outcome; in the pure-state case, perfect distinguishability for a single basis is equivalent to that basis diagonalizing \(\rho_A\) [2003.02103].

## 2. Observable algebras and the operator-algebraic formulation

The central structural object is the observable algebra generated by a family of PVMs,
\[
C^*(K_1,\dots,K_n)\subset B(H_A),
\]
the smallest unital \(*\)-subalgebra containing all spectral projections. If
\[
C^*(K_1,\dots,K_n)=:\mathcal A\subsetneq B(H_A),
\]
then the measurements probe only a proper subalgebra. The missing part of \(B(H_A)\) encodes degrees of freedom that are never tested, physically corresponding to hidden sectors unresolved by coarse-grained measurements. Any finite-dimensional unital \(*\)-subalgebra \(\mathcal A\subset B(H_A)\) admits an Artin-Wedderburn decomposition
\[
\mathcal A \cong \bigoplus_{a=1}^r (M_{n_a}(\mathbb C)\otimes I_{m_a}),
\]
after a unitary identification
\[
H_A \cong \bigoplus_{a=1}^r (\mathbb C^{n_a}\otimes \mathbb C^{m_a}).
\]
Here \(M_{n_a}(\mathbb C)\) is the visible irreducible action, while \(\mathbb C^{m_a}\) is invisible to \(\mathcal A\) [2603.23036].

To encode the memory-assisted structure, fix an orthonormal basis on \(H_A\) and define
\[
\Lambda(X):=\operatorname{Tr}_A[(X^T\otimes I_B)\rho],
\qquad \Lambda(I_A)=\rho_B,
\]
with \(S:=\operatorname{supp}(\rho_B)\subset H_B\), and normalize
\[
\Phi(X):=\rho_B^{-1/2}\Lambda(X)\rho_B^{-1/2},
\qquad \Phi(I_A)=I_S.
\]
For each spectral projection \(P_\alpha\), the ZUS conditions imply that \(\Phi(P_\alpha)\) is a projection and \(\{\Phi(P_\alpha)\}_\alpha\) is a PVM on \(S\). Writing \(\mathcal A^T\) for the transpose image of \(\mathcal A\), the restriction
\[
\phi:=\Phi|_{\mathcal A^T}:\mathcal A^T\to B(S)
\]
satisfies a central equivalence:
\[
\rho \text{ is a common ZUS for all PVMs in } \mathcal A
\iff
\phi \text{ is a unital } *\text{-homomorphism and } \phi(\mathcal A^T)\subset\{\rho_B\}'.
\]
Thus the zero-uncertainty condition is reformulated as multiplicativity on the observable algebra together with a commutant constraint on the memory state [2603.23036].

## 3. Rigidity in the equal-dimension full-algebra regime

The strongest rigidity occurs when the generated observable algebra is the full matrix algebra and the two local Hilbert spaces have equal dimension. If \(H_A\simeq H_B\), \(\rho\) is a common ZUS for a family \(\mathcal K\), and
\[
C^*(\mathcal K)=B(H_A),
\]
then \(\rho\) is pure. The proof uses the multiplicative domain
\[
MD(\Phi)=\{X:\Phi(X^*X)=\Phi(X)^*\Phi(X),\ \Phi(XX^*)=\Phi(X)\Phi(X)^*\},
\]
which becomes all of \(B(H_A)\) when the ZUS conditions force projections in a generating family to map to projections. Hence \(\Phi\) is a unital \(*\)-homomorphism on the full algebra. In finite dimensions this implies that \(\Phi\) is a \(*\)-isomorphism onto \(B(S)\), so by Skolem-Noether there exists a unitary \(U:H_A\to S\) such that
\[
\Phi(X)=UXU^*,
\qquad
\Lambda(X)=(\rho_B^{1/2}U)X(\rho_B^{1/2}U)^*.
\]
Therefore \(\Lambda\) has Kraus rank \(1\), the Choi-Jamiołkowski operator has rank \(1\), and \(\rho\) has rank \(1\) [2603.23036].

Under the same assumptions, purity implies maximal entanglement. Orthogonality of the conditional operators across each generating PVM yields commutation of \(\rho_A\) with all spectral projections, and because the family generates \(B(H_A)\), one obtains
\[
\rho_A\in B(H_A)'=\mathbb CI_A.
\]
Since \(\operatorname{Tr}\rho_A=1\), this forces
\[
\rho_A=\frac{I_A}{d},\qquad d=\dim H_A.
\]
For a pure bipartite state, \(\rho_A=I_A/d\) is equivalent to maximal entanglement, so up to local unitaries
\[
|\psi\rangle=\frac{1}{\sqrt d}\sum_{j=1}^d |j\rangle_A\otimes |j\rangle_B.
\]
This reproduces, in operator-algebraic form, the earlier generic equivalence between zero uncertainty and maximal entanglement established for irreducible sets of nondegenerate observables [2603.23036][2003.02103].

## 4. Failure of rigidity: proper subalgebras and enlarged memory

The full-algebra equal-dimension theorem is not universal. The 2026 analysis identifies two distinct failure mechanisms. The first is the use of a proper observable subalgebra \(\mathcal A\subsetneq B(H_A)\). In that case there exist pure states that are common ZUS for all PVMs contained in \(\mathcal A\) but are not globally maximally entangled. Using the decomposition
\[
H_A \cong \bigoplus_{a=1}^r (\mathbb C^{n_a}\otimes \mathbb C^{m_a}),\qquad
\mathcal A \cong \bigoplus_{a=1}^r (B(\mathbb C^{n_a})\otimes I_{m_a}),
\]
one may choose a block \(a_0\) and construct
\[
|\psi\rangle:=|\Phi_{n_{a_0}}\rangle_{A_{a_0,1}B_{a_0,1}}\otimes |\eta\rangle_{A_{a_0,2}B_{a_0,2}},
\]
where \(|\Phi_n\rangle\) is maximally entangled on the visible factor and \(|\eta\rangle=|u\rangle\otimes |v\rangle\) is a product state on the multiplicity part. Then the conditional operators for any PVM \(\{E_\alpha\}\subset \mathcal A\) take the form
\[
Z_\alpha=\frac{1}{n_{a_0}}(E_\alpha^{(a_0)})^T\otimes |v\rangle\langle v|,
\]
which are pairwise orthogonal, while
\[
\rho_A=\frac{1}{n_{a_0}}I_{n_{a_0}}\otimes |u\rangle\langle u|
\]
is not maximally mixed on \(H_A\) [2603.23036].

The second mechanism is larger memory dimension. If \(\dim H_A=d\), \(\dim H_B\ge d\), and \(\rho\) is a common ZUS for a family generating \(B(H_A)\), then there exist a finite-dimensional \(K\), a unitary \(U:S\to \mathbb C^d\otimes K\), and a density operator \(\sigma\in B(K)\) such that
\[
U\phi(X)U^*=X\otimes I_K,\qquad
U\rho_BU^*=\frac{1}{d}I_d\otimes \sigma,\qquad
U\Lambda(X)U^*=\frac{1}{d}X\otimes \sigma.
\]
Equivalently, up to local isometries,
\[
(I_A\otimes U)\rho(I_A\otimes U^*)=|\Phi_d^+\rangle\langle \Phi_d^+|_{AB_1}\otimes \sigma_{B_2}.
\]
The visible sector is maximally entangled on a \(d\otimes d\) subsystem, but the extra memory factor is unconstrained. This is the operator-algebraic analogue of the “subsystem maximal entanglement plus ancilla” structure already present in the nondegenerate theory [2603.23036][2003.02103].

More generally, if
\[
\mathcal A^T \cong \bigoplus_{a=1}^r (M_{n_a}(\mathbb C)\otimes I_{m_a}),
\]
and \(\phi:\mathcal A^T\to B(S)\) is a unital \(*\)-homomorphism with \(\phi(\mathcal A^T)\subset\{\rho_B\}'\), then there exist finite-dimensional spaces \(K_a\) and a unitary
\[
U:S\to \bigoplus_{a=1}^r (\mathbb C^{n_a}\otimes K_a)
\]
such that
\[
U\phi(X)U^*=\bigoplus_{a=1}^r (X_a\otimes I_{K_a}),
\]
\[
U\rho_BU^*=\bigoplus_{a=1}^r (I_{n_a}\otimes \tau_a),
\]
\[
U\Lambda_{\mathcal A^T}(X)U^*=\bigoplus_{a=1}^r (X_a\otimes \tau_a),
\]
with \(\tau_a\ge 0\) on \(K_a\). Zero uncertainty therefore determines only the visible matrix factors; multiplicity spaces and commuting memory states encode the unconstrained degrees of freedom [2603.23036].

## 5. Entropic uncertainty, irreducibility, and prior formulations

The operator-algebraic results are closely tied to memory-assisted entropic uncertainty relations. For two projective measurements \(X\) and \(Z\) on \(H_A\),
\[
H(X|B)+H(Z|B)\ge \log(1/c)+H(A|B),
\]
where \(c=\max_{i,j}|\langle x_i|z_j\rangle|^2\). If both conditional entropies vanish, then
\[
0\ge \log(1/c)+H(A|B).
\]
For mutually unbiased bases in dimension \(d\), \(c=1/d\), so
\[
H(A|B)\le -\log d.
\]
For pure states, \(H(A|B)=-S(\rho_A)\), and \(S(\rho_A)=\log d\) only for maximal entanglement. This entropic argument matches the full-algebra rigidity theorem in the equal-dimension case [2603.23036].

The earlier classification for nondegenerate observables is organized by the transition graph \(G(\mathcal B)\) of the measurement bases. If the basis set is irreducible, then a bipartite state is a ZUS if and only if \(d_B\ge d=\dim H_A\) and the state is maximally entangled. If the basis set is reducible, then ZUS decompose into direct sums of component maximally entangled states with orthogonal supports on \(B\), and the minimum entanglement required is \(\log r_{\min}\), where \(r_{\min}\) is the smallest component rank [2003.02103]. The 2026 framework subsumes this by replacing transition-graph irreducibility with the observable algebra \(C^*(\mathcal K)\) and by explicitly incorporating degenerate PVMs and commutants [2603.23036].

The entropic perspective also admits refinements. A strengthened lower bound includes an additional nonnegative term depending on the quantum discord and the classical correlations of \(\rho_{AB}\),
\[
S(P|B)+S(Q|B)\ge -2\log_2 c(P,Q)+S(A|B)+\max\{0,D_A(\rho_{AB})-J_A(\rho_{AB})\},
\]
so whenever \(D_A(\rho_{AB})>J_A(\rho_{AB})\), the lower bound is strictly tighter than the Berta-type bound. This further constrains the possibility of zero uncertainty for mixed or non-maximally entangled states [1204.3803]. Experimentally, entanglement-assisted reduction of conditional uncertainty was realized with entangled photon pairs, an optical delay line serving as a quantum memory, and fast feed-forward; for complementary qubit observables the observed conditional uncertainties approached the ideal zero-uncertainty limit for Bell states [1012.0332].

## 6. Examples, tests, and scope

A representative degenerate example is the qutrit maximally entangled state
\[
|\Phi_3\rangle=\frac{1}{\sqrt 3}\sum_{j=0}^2 |jj\rangle,\qquad \rho=|\Phi_3\rangle\langle\Phi_3|,
\]
with binary PVMs
\[
\mathcal P=\{P_0,P_1\},\quad P_0=|0\rangle\langle 0|+|1\rangle\langle 1|,\quad P_1=|2\rangle\langle 2|,
\]
and
\[
\mathcal Q=\{Q_0,Q_1\},\quad Q_1=|v\rangle\langle v|,\quad Q_0=I-|v\rangle\langle v|,\quad |v\rangle=\frac{|1\rangle+|2\rangle}{\sqrt 2}.
\]
For maximally entangled \(\rho\),
\[
\operatorname{Tr}_A[(X\otimes I)\rho]=X^T/3,
\]
so the assemblages are
\[
\sigma_{a|\mathcal P}=P_a^T/3,\qquad \sigma_{b|\mathcal Q}=Q_b^T/3.
\]
For each setting,
\[
\sigma_{0|\cdot}\sigma_{1|\cdot}=0,
\]
and Bob can identify Alice’s coarse-grained outcome with zero error once the setting is revealed. This is perfect coarse-grained steering in a degenerate-measurement setting [2603.23036].

The same paper gives explicit proper-subalgebra and larger-memory examples. If
\[
\mathcal A=B(A_1)\otimes I_{A_2}\subsetneq B(H_A),\qquad
\rho=|\Phi_2^+\rangle\langle \Phi_2^+|_{A_1B_1}\otimes \omega_{A_2B_2},
\]
then for any PVM \(\{E_\alpha\}\subset B(A_1)\),
\[
\operatorname{Tr}_A[((E_\alpha\otimes I_{A_2})\otimes I_B)\rho]=\frac{1}{2}E_\alpha^T\otimes \tau,
\qquad \tau=\operatorname{Tr}_{A_2}\omega,
\]
so the conditional operators are orthogonal although \(\rho_A\) is not maximally mixed on \(H_A\). If instead
\[
\rho=|\Phi_2^+\rangle\langle \Phi_2^+|_{AB_1}\otimes \sigma_{B_2}
\]
with full observable algebra on \(A\), then for any PVM \(\{E_\alpha\}\),
\[
\operatorname{Tr}_A[(E_\alpha\otimes I_B)\rho]=\frac{1}{2}E_\alpha^T\otimes \sigma,
\]
so common ZUS holds while the global state is pure only when \(\sigma\) is pure. These examples display the two failure mechanisms in concrete form [2603.23036].

The operator-algebraic criteria also yield a practical test for whether \(\rho_{AB}\) is an \(\mathcal A\)-ZUS. First compute the conditional operators \(Z_\alpha\) for generating PVMs and verify
\[
Z_\alpha Z_\beta=0\quad (\alpha\neq\beta),\qquad \sum_\alpha Z_\alpha=\rho_B.
\]
Then compute
\[
\Phi(X)=\rho_B^{-1/2}\operatorname{Tr}_A[(X^T\otimes I)\rho]\rho_B^{-1/2},
\]
restrict \(\phi=\Phi|_{\mathcal A^T}\), and check that \(\phi\) is a unital \(*\)-homomorphism on generators and that
\[
\phi(\mathcal A^T)\subset \{\rho_B\}'.
\]
Finally, diagonalize the representation to obtain the normal form
\[
U\phi(X)U^*=\bigoplus_a (X_a\otimes I_{K_a}),\qquad
U\rho_BU^*=\bigoplus_a (I_{n_a}\otimes \tau_a),
\]
which identifies the visible blocks and the unconstrained multiplicity data [2603.23036].

The present framework is finite-dimensional and uses finite-dimensional \(C^*\)-algebra machinery, including Artin-Wedderburn decomposition, Stinespring dilation, and multiplicative-domain methods. Infinite-dimensional or continuous-variable extensions are not addressed, and a complete classification of all \(\mathcal A\)-ZUS at the level of global states \(\rho\), beyond the visible data \((\phi,\rho_B)\), remains open. This suggests that the principal conceptual advance is not merely the extension from rank-one to degenerate measurements, but the identification of observable algebras and their commutants as the precise locus where rigidity, multiplicity, and physical coarse-graining meet [2603.23036].

Source: https://www.emergentmind.com/topics/zero-uncertainty-states-with-quantum-memory