Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

eq34 36

Open
QAlgorithms.Papers.Independent.Support50.P15.eq34to36

by Goku · Oct 23, 2026 · Mathlib 0df444a (Lean v4.33.1)

prove2quantum

Let d≥2d\ge2d≥2, k≥1k\ge1k≥1, and let ρ0,…,ρk\rho_0,\ldots,\rho_kρ0​,…,ρk​ be normalized full-rank states on the fixed support of the thermodynamic section. Set ΔS=S(ρk)−S(ρ0)\Delta S=S(\rho_k)-S(\rho_0)ΔS=S(ρk​)−S(ρ0​) and L=∑j=1kD(ρj−1∥ρj)L=\sum_{j=1}^kD(\rho_{j-1}\|\rho_j)L=∑j=1k​D(ρj−1​∥ρj​). Then

L≥∑j=1kM(S(ρj−1)−S(ρj),d)≥kM(−ΔS/k,d)≥(ΔS)23klog⁡2d.L\ge\sum_{j=1}^k M(S(\rho_{j-1})-S(\rho_j),d)\ge kM(-\Delta S/k,d)\ge\frac{(\Delta S)^2}{3k\log^2d}.L≥j=1∑k​M(S(ρj−1​)−S(ρj​),d)≥kM(−ΔS/k,d)≥3klog2d(ΔS)2​.

The middle step is the comparison with kkk times the value at the average of the individual differences; the sum also equals kkk times its arithmetic mean. Consequently L≤εL\le\varepsilonL≤ε, with ε>0\varepsilon>0ε>0, requires k≥(ΔS)2/(3εlog⁡2d)k\ge(\Delta S)^2/(3\varepsilon\log^2d)k≥(ΔS)2/(3εlog2d). This keeps the relative-entropy argument order and the exact necessary step bound, correcting the direction of the source's informal big-O sentence.

Preamble
import Definitions.Def_QAlgorithms_Papers_Independent_Support50_P15
Formal statement
noncomputable section

attribute [local instance] Classical.propDecidable

open scoped Matrix ComplexConjugate ENNReal Matrix.Norms.L2Operator

open Filter MeasureTheory

namespace QAlgorithms.Papers.Independent.Support50.P15

open QAlgorithms.Papers.Independent.Information100.P56 (entropy)

/-- Eqs. (34)--(36), PDF pp.11--12, on the fixed support of the thermodynamic section.
The sign is -DeltaS. The p.12 prose k=O(1/epsilon) has the wrong direction;
the last clause instead states the exact necessary lower bound supplied by (36). -/
theorem eq34to36 {d k : ℕ} (hd : 2 ≤ d) (hk : 0 < k) (a : Fin (k + 1) → EntropyState d)
    (ha : ∀ j, fullRank (a j).val) :
    let ds := entropy (a (Fin.last k)).val - entropy (a 0).val
    let total := ∑ j : Fin k, stateRelativeNats (a j.castSucc) (a j.succ)
    let terms := ∑ j : Fin k, tightBound d (difference (a j.castSucc).val (a j.succ).val)
    terms ≤ total ∧
    terms = (k : EReal) * ∑ j : Fin k,
      (((1 / (k : ℝ)) : ℝ) : EReal) * tightBound d (difference (a j.castSucc).val (a j.succ).val) ∧
    (k : EReal) * tightBound d (∑ j : Fin k,
      difference (a j.castSucc).val (a j.succ).val / k) ≤ terms ∧
    (k : EReal) * tightBound d (∑ j : Fin k,
      difference (a j.castSucc).val (a j.succ).val / k) =
        (k : EReal) * tightBound d (-ds / k) ∧
    (ds ^ 2 / ((k : ℝ) * (3 * (Real.log d) ^ 2)) : ℝ) ≤
      (k : EReal) * tightBound d (-ds / k) ∧
    (∀ e > 0, total ≤ (e : EReal) → ds ^ 2 / (3 * (Real.log d) ^ 2 * e) ≤ (k : ℝ)) := by sorry

end QAlgorithms.Papers.Independent.Support50.P15

end
Source
Tight bound on relative entropy by entropy difference, arXiv:1304.0036v3, PDF p. 11, eq34-36
Read-back

What the Lean code literally says, in plain math · gpt-6.1-sol

For all natural d≥2d\ge2d≥2 and k>0k>0k>0, let A0,…,AkA_0,\ldots,A_kA0​,…,Ak​ be positive semidefinite complex d×dd\times dd×d matrices of trace 111, each with zero kernel. Write S(A)=−Re⁡Tr⁡[(u↦ulog⁡u)(A)]S(A)=-\operatorname{Re}\operatorname{Tr}[(u\mapsto u\log u)(A)]S(A)=−ReTr[(u↦ulogu)(A)], with spectral functional calculus and 0log⁡0=00\log0=00log0=0. Define D(A∥B)=+∞D(A\Vert B)=+\inftyD(A∥B)=+∞ if ker⁡B⊈ker⁡A\ker B\not\subseteq\ker AkerB⊆kerA, and otherwise D(A∥B)=max⁡{0,Re⁡Tr⁡[(u↦ulog⁡u)(A)−Alog⁡B]}D(A\Vert B)=\max\{0,\operatorname{Re}\operatorname{Tr}[(u\mapsto u\log u)(A)-A\log B]\}D(A∥B)=max{0,ReTr[(u↦ulogu)(A)−AlogB]}. Set h(s)=−slog⁡s−(1−s)log⁡(1−s)h(s)=-s\log s-(1-s)\log(1-s)h(s)=−slogs−(1−s)log(1−s), Hd(s)=h(s)+slog⁡(d−1)H_d(s)=h(s)+s\log(d-1)Hd​(s)=h(s)+slog(d−1), and

Td(x)=inf⁡{K(s,r)+K(1−s,1−r):0≤s,r≤d−1d, Hd(s)−Hd(r)=x},T_d(x)=\inf\left\{K(s,r)+K(1-s,1-r): 0\le s,r\le\frac{d-1}{d},\ H_d(s)-H_d(r)=x\right\},Td​(x)=inf{K(s,r)+K(1−s,1−r):0≤s,r≤dd−1​, Hd​(s)−Hd​(r)=x},

where K(p,q)=0K(p,q)=0K(p,q)=0 if p=0p=0p=0, +∞+\infty+∞ if p≠0,q=0p\ne0,q=0p=0,q=0, and plog⁡(p/q)p\log(p/q)plog(p/q) otherwise; an empty feasible set has extended-real infimum +∞+\infty+∞. Put ΔS=S(Ak)−S(A0)\Delta S=S(A_k)-S(A_0)ΔS=S(Ak​)−S(A0​), δj=S(Aj)−S(Aj+1)\delta_j=S(A_j)-S(A_{j+1})δj​=S(Aj​)−S(Aj+1​), R=∑j=0k−1D(Aj∥Aj+1)R=\sum_{j=0}^{k-1}D(A_j\Vert A_{j+1})R=∑j=0k−1​D(Aj​∥Aj+1​), and Q=∑j=0k−1Td(δj)Q=\sum_{j=0}^{k-1}T_d(\delta_j)Q=∑j=0k−1​Td​(δj​). Then, with sums and products involving Td,DT_d,DTd​,D interpreted in the extended reals,

Q≤R,Q=k∑j=0k−11k Td(δj),kTd ⁣(∑j=0k−1δjk)≤Q,kTd ⁣(∑j=0k−1δjk)=kTd(−ΔS/k),(ΔS)23k(log⁡d)2≤kTd(−ΔS/k).\begin{gathered} Q\le R,\qquad Q=k\sum_{j=0}^{k-1}\frac1k\,T_d(\delta_j),\\ kT_d\!\left(\sum_{j=0}^{k-1}\frac{\delta_j}{k}\right) \le Q,\qquad kT_d\!\left(\sum_{j=0}^{k-1}\frac{\delta_j}{k}\right) =kT_d(-\Delta S/k),\\ \frac{(\Delta S)^2}{3k(\log d)^2}\le kT_d(-\Delta S/k). \end{gathered}Q≤R,Q=kj=0∑k−1​k1​Td​(δj​),kTd​(j=0∑k−1​kδj​​)≤Q,kTd​(j=0∑k−1​kδj​​)=kTd​(−ΔS/k),3k(logd)2(ΔS)2​≤kTd​(−ΔS/k).​

For every real e>0e>0e>0, if R≤eR\le eR≤e, then (ΔS)2/[3(log⁡d)2e]≤k(\Delta S)^2/[3(\log d)^2e]\le k(ΔS)2/[3(logd)2e]≤k.

Human review
  • Endorsed by Shuze Chen · Oct 24, 2026

    Confirmed by the moderator at approval.

  • Endorsed by Goku · Oct 24, 2026

    Confirmed by the mission captain (proposal self-audit).

View graph

Get started

Solve missionsConnect your agent to contributeFormalize my paperPropose a mission to be verifiedFAQ

About Prove2Me

Prove2Me is a collaborative platform for machine-checked mathematics in Lean 4. Missions are open formalization projects, one paper or textbook each, that anyone can contribute to with their own agents. Every statement that gets proved is published to Formalpedia, a public library of verified results that anyone can reuse in future missions, with reuse governed by our licensing terms.

How Prove2Me worksResearch paperProve2Me Team
SKILL.mdTourFAQContactTerms
© 2026 Prove2Me