RESEARCH / Focused target
Non-isotropic mixed-unitary qubit channels do not commute with marginal pinching
RELEASED FOUNDATIONS / PROPOSED CONNECTION
Research connections
All source anchors (1)
- D5/S3/Quantum/Information/MixedUnitaryPinchingCommutation.resultAbsent from this graph
Problem
The question
Z.-W. Liu, R. Takagi and S. Lloyd, Diagonal quantum discord, J. Phys. A: Math. Theor. 52, 135301 (2019), arXiv:1708.09076v2, Section 3.2, “$\pi$-commuting local operations and monotonicity”, conjecture following the probabilistic-Hadamard example:
We conjecture (which is not important for our current purpose) that ${\bf ISO} = X_A(\pi_A)$. That is, qubit mixed-unitary channels that are not isotropic all fail to satisfy the condition.
This quotation is line 463 of the source archive's revised_jphysa_clean.tex (md5 3ae0def434187ebe1875d87961e23048). The relevant definitions are at lines 261, 325–329 and 347; line 339 assumes the input marginal nondegenerate. See also the literature note.
The formal definitions use the following conventions:
IsMixedUnitarymeans a finite mixture $\Phi(a)=\sum_j p_j U_j a U_j^\dagger$ on every complex $2\times2$ matrix, with real $p_j\ge0$, $\sum_jp_j=1$, and unitary $U_j$.IsIsotropicmeans that, for some fixed $\gamma\in[0,1]$ and unitary $U$, either $\Phi(a)=(1-\gamma)UaU^\dagger+\gamma\operatorname{tr}(a)I/2$ for every $a$, or $\Phi(a)=(1-\gamma)Ua^{\mathsf T}U^\dagger+\gamma\operatorname{tr}(a)I/2$ for every $a$. Transpose gives the complex linear extension of the antiunitary action on Hermitian states.localMapapplies $\Phi$ to each first-qubit block; for these linear maps it is $\Phi\otimes\mathrm{id}_B$. The marginal $\rho_A=\operatorname{tr}_B\rho$ ispartialTraceRight.IsNondegeneratemeans Hermitian with two distinct eigenvalues. For a Hermitian marginal,pinchAis $\pi_A(\rho)=\sum_i(P_i\otimes I)\rho(P_i\otimes I)$, where $P_i$ are its rank-one spectral projectors; its spectral basis is recomputed on the argument to which pinching is applied.
Issue #22787 fixes the reading to the quotation's second sentence: for every mixed-unitary qubit map $\Phi$ that is not isotropic, there is a positive semidefinite two-qubit matrix $\rho$ of trace one such that $\rho_A$ and $\Phi(\rho_A)$ are nondegenerate and $\pi_A((\Phi\otimes\mathrm{id}_B)(\rho))\ne(\Phi\otimes\mathrm{id}_B)(\pi_A(\rho))$. This is proved.
Motivation
Our foothold
The source uses commutation with the discord-destroying map as a sufficient condition for monotonicity of relative-entropy diagonal discord. Classifying the mixed-unitary channels satisfying that condition identifies the reach of this monotonicity argument. Failure to commute does not by itself assert an increase of diagonal discord.
Gap
Missing bridges
The literature check before any Lean is recorded in #22787:
- the conjecture occurs in the 2019 arXiv v2 source;
- the oracle-reported searches of Zenodo, ResearchGate, SSRN and GitHub by title and conjecture wording found no settlement;
- the oracle-reported readings of Sone et al., Phys. Rev. A 98, 012115 (2018), and Phys. Rev. A 99, 052318 (2019), found no settlement of this classification;
- the orchestrator's searches of repository issues and of
Problems/,Library/andD5/found no prior record of the problem.
not-found-in-searched-scope.
Route
Proposed approach
Write $S(v)=v\cdot\sigma$ for the real Pauli combination, $T$ for blochOperator, and $Q=T^{\mathsf T}T$.
- Bloch representation. For a qubit map satisfying
IsMixedUnitary,mixed_channel_datasupplies a complex linear representative, preservation of the identity, trace and Hermiticity, and $\Phi(S(v))=S(Tv)$ for every real $v\in\mathbb R^3$. Here $T_{ij}=\operatorname{Re}\operatorname{tr}(\sigma_{i+1}\Phi(\sigma_{j+1}))/2$. - Scalar Gram matrices. For a mixed-unitary $\Phi$ and real $c$ with $T^{\mathsf T}T=cI$,
scalar_gram_boundsgives $0\le c\le1$. For any real $3\times3$ matrix $O$ with $O^{\mathsf T}O=I$,orthogonal_spin_liftimplements its action on all Pauli combinations by either unitary conjugation or unitary conjugation after transpose. Thusscalar_gram_isotropic, assumingIsMixedUnitaryand the existence of such a real $c$, givesIsIsotropic: $c=0$ gives complete depolarization, while $c>0$ uses $T=\sqrt c,O$ and $\gamma=1-\sqrt c$. Non-isotropy therefore excludes a scalar Gram matrix. - Orthogonal probes. For any real $3\times3$ matrix $T$ for which no real $c$ satisfies $T^{\mathsf T}T=cI$,
orthogonal_probesupplies real $n,x$ with $n\cdot n=x\cdot x=1$, $n\cdot x=0$ and $(Tn)\cdot(Tx)\ne0$. A nonzero off-diagonal entry of $Q$ supplies coordinate directions; otherwise unequal diagonal entries supply their sum and difference.normalize_probeapplies to real $a,b$ with $a\cdot b=0$ and $(Ta)\cdot(Tb)\ne0$, and normalizes them without losing either condition. - A density matrix and its image. Set $R(n,x)=\tfrac14[I\otimes I+\tfrac12S(n)\otimes I+\tfrac14S(x)\otimes Z]$, the definition
witness. Under $n\cdot n=x\cdot x=1$ and $n\cdot x=0$,witness_densityproves positive semidefiniteness and trace one. For arbitrary real $n,x$,witness_marginalgives $\operatorname{tr}_B R(n,x)=(I+S(n)/2)/2$. For mixed-unitary $\Phi$ and arbitrary real $n,x$,local_witnessgives $(\Phi\otimes\mathrm{id}_B)R(n,x)=R(Tn,Tx)$, andmapped_marginalidentifies its marginal with $\Phi(\operatorname{tr}_B R(n,x))=(I+S(Tn)/2)/2$. - Spectral pinching. For any complex two-qubit matrix $\tau$ whose first marginal is $(I+S(v))/2$ with real $v\ne0$,
pinching_formulagives $\pi_A(\tau)=\tfrac12[\tau+(v\cdot v)^{-1}(S(v)\otimes I)\tau(S(v)\otimes I)]$. Consequentlypinch_witness, for arbitrary real $a,b$ with $a\ne0$, gives $\pi_A(R(a,b))=R(a,[(a\cdot b)/(a\cdot a)]a)$. For real $v\ne0$,bloch_nondegenerateproves that $(I+S(v))/2$ is nondegenerate; applying it to $n/2$ and $Tn/2$ supplies the two required nondegeneracy conclusions. - Nonzero commutator. For mixed-unitary $\Phi$, real $n,x$ with $n\cdot n=1$, $n\cdot x=0$, and $Tn\ne0$,
commutator_identitygives $\pi_A((\Phi\otimes\mathrm{id}_B)R(n,x))-(\Phi\otimes\mathrm{id}_B)(\pi_A(R(n,x)))=\frac{(Tn)\cdot(Tx)}{16,(Tn)\cdot(Tn)}S(Tn)\otimes Z$. For any real $a,b$ with $a\cdot b\ne0$,commutator_nonzeroproves that the corresponding expression $\frac{a\cdot b}{16,a\cdot a}S(a)\otimes Z$ is nonzero. The probes satisfy these hypotheses, yieldingresult.
Falsifier
What would falsify this route
The settlement would miss the registered problem if the channel class, marginal or pinching differed from the source's. The definitions use actual finite unitary mixtures, both isotropic forms, the partial trace over the second qubit, and the first marginal's spectral projectors. The antiunitary alternative is not omitted.
pinchA chooses a spectral basis even at a degenerate Hermitian marginal and returns its argument outside the Hermitian-marginal domain. result avoids these extensions: both $\rho_A$ and $\Phi(\rho_A)$ are nondegenerate, and mapped_marginal, for mixed-unitary $\Phi$ and arbitrary real $n,x$, identifies the latter with the actual output marginal of $R(n,x)$. The probes have $(Tn)\cdot(Tx)\ne0$, which forces $Tn\ne0$, so the commutator formula uses no inverse at zero. The zero Bloch matrix belongs to the scalar-Gram, isotropic case. Invertibility of $T$ is not assumed.
Evidence
Evidence to collect
The canonical source is D5/S3/Quantum/Information/MixedUnitaryPinchingCommutation.lean.
- Proof. It proves
resultwith the universally quantified existence statement in Problem. - Axioms. It uses only the standard axioms
propext,Classical.choiceandQuot.sound; nosorry,native_decide, or new axiom.
Frozen state: Golden/Frozen/state/D5/S3/Quantum/Information/MixedUnitaryPinchingCommutation.lean.json.
Declaration statements:
result:sha256:99836e9c6706852219dcffb2afe124e67e66365bba871d411be5c464e596b7ae.
Triage
Scope assessment
Tier 1 conjecture of a 2019 journal paper, preregistered in issue #22787 before any Lean.
theorem; resolution proved.
The module is admitted under the open-problem-resolution basis of CLAUDE.md §3.2; the per-declaration proof shapes are listed in the delivery PR. There is no digestion atom.
What the settlement shows
Proved by result: for every map on complex qubit matrices satisfying IsMixedUnitary and not IsIsotropic, there exists a positive semidefinite two-qubit matrix of trace one whose first marginal and its image under the map are both nondegenerate, and on which marginal pinching and the local map do not commute.
Computed, not formalized: the random mixed-unitary channels and random states that the orchestrator recorded in #22787 agree with the formal statement: each constructed witness for a non-isotropic channel gave a nonzero commutator, and no violation was observed for the tested isotropic channel.
Argued, not formalized.
- Decisive mechanism and singular maps. For mixed-unitary $\Phi$ and unit orthogonal $n,x$ with $(Tn)\cdot(Tx)\ne0$,
local_witnessmaps $R(n,x)$ to $R(Tn,Tx)$;pinch_witness, applicable to any real $a\ne0$ and $b$, retains only the component of $b$ parallel to $a$. Thus the correlation along $x$ is erased at the input, while its transformed component parallel to $Tn$ survives output pinching. The nonzero inner product guarantees $Tn\ne0$ even when $T$ is singular. This mechanism uses no lower bound on a singular value;resultsupplies no quantitative lower bound on the commutator. - Entanglement is unnecessary for this construction. For unit orthogonal $n,x$, write $\tau_\pm=[I+S(n)/2\pm S(x)/4]/2$. Their Bloch vectors have length $\sqrt5/4<1$, so they are qubit density matrices, and $R(n,x)=\tfrac12\tau_+\otimes|0\rangle\langle0|+\tfrac12\tau_-\otimes|1\rangle\langle1|$ is separable. Each $\tau_\pm$ has eigenvalues $(4\pm\sqrt5)/8$, so $R(n,x)$ has the positive eigenvalues $(4+\sqrt5)/16$ and $(4-\sqrt5)/16$, each twice. These separability and exact-spectrum statements are additional arguments, not conclusions stated by
result. - Why the mixed-unitary restriction matters. Fix a nondegenerate qubit density $\sigma=\sum_i p_i|i\rangle\langle i|\ne I/2$ and let $\Phi_\sigma(a)=\operatorname{tr}(a)\sigma$. The Kraus operators $\sqrt{p_i}|i\rangle\langle j|$, for $i,j\in{0,1}$, give this channel and satisfy $\sum_{i,j}K_{ij}^\dagger K_{ij}=I$. Its local output is $\sigma\otimes\rho_B$, fixed by pinching in the eigenbasis of $\sigma$. Input pinching preserves $\rho_B$, since the input spectral projectors sum to the identity, so both orders give $\sigma\otimes\rho_B$ for every bipartite density with nondegenerate input marginal. But $\Phi_\sigma(I)=2\sigma\ne I$, whereas every mixed-unitary or isotropic map defined above is unital. This supplies commuting channels outside those classes and shows why the quotation's unrestricted set equality needs qualification. The formal settlement is its mixed-unitary sentence.
- Effect on the paper and neighbouring questions. The source's isotropic inclusion (TeX line 412) and its qubit monotonicity corollary for $\mathbf{SC}\cup\mathbf{ISO}$ (line 454) do not depend on the conjecture at line 463. On the nondegenerate input-and-output domain, the settlement supplies the complementary exclusion of every non-isotropic mixed-unitary channel; the source's converse is not formalized here. Its separate Conjecture 1 (line 470), “For $d_A= 2$, diagonal discord is monotonically nonincreasing under any local discord nongenerating channel.”, remains open in this settlement: noncommutation removes a sufficient monotonicity argument, but gives no entropy comparison. No higher-dimensional classification or universal treatment of degenerate marginals is supplied by
result.
ASSUMED-UNVERIFIED
Unverified assumptions
The literature check is bounded and does not establish exhaustive worldwide novelty or priority. The v2 TeX source and its conjecture were independently read; the preregistration's arXiv revision metadata, Crossref reading, AQIS 2017 slides claim, prior-settlement web searches, and readings of Sone et al., Phys. Rev. A 98, 012115 (2018), and Phys. Rev. A 99, 052318 (2019), were not independently reproduced.
The numerical checks are the orchestrator's readings in #22787 and were not rerun. All statements labelled “Computed, not formalized” or “Argued, not formalized”, including separability, the exact witness spectrum, the replacement-channel example and the implications for the paper, are outside the statement of result. The published isotropic converse and monotonicity results remain external literature results; the separate monotonicity conjecture, higher-dimensional extensions and degenerate-marginal classification are not settled here.