OctoInformBeta

OpenAI preprint, September 2026 · 4. From the thermal inequality to channel outputs · Section 4

Corollary 4.4: A lower bound for mixed and correlated inputs under tensor powers of a generalized amplitude-damping channel

Output entropy from local input populations.

Hover over a blue or green term in the text, or a node in the graph, to find it in the other. Select it to read its declaration both in Lean and in words.

The Setting

A block of nn qubits, with nn a nonnegative integer, is sent through independent copies of one generalized amplitude-damping channel. Its input may be mixed and correlated.

The computational basis ℬn\mathcal B_n consists of bit strings Fin⁡(n)→Fin⁡(2)\operatorname{Fin}(n)\to\operatorname{Fin}(2), where Fin⁡(m)={0,…,m−1}\operatorname{Fin}(m)=\{0,\ldots,m-1\}.✦ An nn-qubit operator is a complex matrix with rows and columns indexed by ℬn\mathcal B_n.✦ A density operator PP on nn qubits is a complex matrix with rows and columns indexed by ℬn\mathcal B_n that is positive semidefinite and has trace one.✦

For damping γ∈[0,1]\gamma\in[0,1] and stationary excited-state population ν∈[0,1]\nu\in[0,1], the single-qubit Kraus operators Kr(1)K^{(1)}_r, indexed by r∈Fin⁡(4)r\in\operatorname{Fin}(4), are given by (1): K0(1)=1−ν(1001−γ),K1(1)=γ(1−ν)(0100),K2(1)=ν(1−γ001),K3(1)=γν(0010).\begin{aligned} K^{(1)}_0&=\sqrt{1-\nu} \begin{pmatrix}1&0\\0&\sqrt{1-\gamma}\end{pmatrix}, &K^{(1)}_1&=\sqrt{\gamma(1-\nu)} \begin{pmatrix}0&1\\0&0\end{pmatrix},\\ K^{(1)}_2&=\sqrt{\nu} \begin{pmatrix}\sqrt{1-\gamma}&0\\0&1\end{pmatrix}, &K^{(1)}_3&=\sqrt{\gamma\nu} \begin{pmatrix}0&0\\1&0\end{pmatrix}. \end{aligned}(1)✦

The tensor product Kraus operator KrK_r, for a label r:Fin⁡(n)→Fin⁡(4)r:\operatorname{Fin}(n)\to\operatorname{Fin}(4), is the nn-qubit operator whose entries at i,k∈ℬni,k\in\mathcal B_n are defined by (2): Kr(i,k)=∏j=0n−1Kr(j)(1)(i(j),k(j)).K_r(i,k)=\prod_{j=0}^{n-1}K^{(1)}_{r(j)}\bigl(i(j),k(j)\bigr).(2)✦ The memoryless channel 𝒩γ,ν⊗n\mathcal N_{\gamma,\nu}^{\otimes n} is the map on nn-qubit operators given by (3), where ** denotes the adjoint: 𝒩γ,ν⊗n(P)=∑r:Fin⁡(n)→Fin⁡(4)KrPKr*.\mathcal N_{\gamma,\nu}^{\otimes n}(P) =\sum_{r:\operatorname{Fin}(n)\to\operatorname{Fin}(4)}K_r P K_r^*.(3)✦

For an nn-qubit operator PP and a site j∈Fin⁡(n)j\in\operatorname{Fin}(n), the local excitation population pj(P)p_j(P) is the sum in (4): pj(P)=∑i∈ℬni(j)=1Re⁡Pii.p_j(P)=\sum_{\substack{i\in\mathcal B_n\\i(j)=1}} \operatorname{Re}P_{ii}.(4)✦

For a finite square complex matrix AA, its entropy S(A)S(A) is the real part of the trace of the spectral function x↦−xln⁡xx\mapsto-x\ln x applied to AA.✦ On a density operator PP, the von Neumann entropy takes the form (5), in nats: S(P)=−Tr⁡(Pln⁡P).S(P)=-\operatorname{Tr}(P\ln P).(5)✦ Natural logarithms and the convention 0ln⁡0=00\ln0=0 are used throughout.

The qubit entropy function gg at determinant uu is defined from binary entropy hh by (6): h(q)=−qln⁡q−(1−q)ln⁡(1−q),g(u)=h(1+1−4u2).\begin{aligned} h(q)&=-q\ln q-(1-q)\ln(1-q),\\ g(u)&=h\!\left(\frac{1+\sqrt{1-4u}}{2}\right). \end{aligned}(6)✦ The output determinant function vv, for an input excitation population p∈[0,1]p\in[0,1], is defined by (7): v(γ,ν,p)=γν(1−ν)+γ(1−γ)(p−ν)2.v(\gamma,\nu,p)=\gamma\nu(1-\nu) +\gamma(1-\gamma)(p-\nu)^2.(7)✦

The Claim

Theorem 1 (Corollary 4.4). Let γ,ν∈[0,1]\gamma,\nu\in[0,1] and let PP be a density operator on nn qubits, for a nonnegative integer nn. The output entropy satisfies (8): ∑j=0n−1g(v(γ,ν,pj(P)))≤S(𝒩γ,ν⊗n(P)).\sum_{j=0}^{n-1}g\bigl(v(\gamma,\nu,p_j(P))\bigr) \leq S\!\left(\mathcal N_{\gamma,\nu}^{\otimes n}(P)\right).(8)✦

What Is Checked

For each nonnegative integer nn and γ,ν∈[0,1]\gamma,\nu\in[0,1], the claim is stated over density operators on ℬn\mathcal B_n, with the channel (3) and the population function (4).