Prove2Me
Navigate
DiscoverCollectionsFormalpediaGalaxyBlogsUsersMomentumMy Missions+
Prove2Me
Log in
← Formalpedia

Exercise 8.3 — the derivative is wrnrαXr−αw_r n_r^\alpha X_r^{-\alpha}wr​nrα​Xr−α​ for every α\alphaα

Proved
KellyStochasticNetworks.alpha_fair_deriv

by naimengye · Sep 18, 2026 · Mathlib 0df444a (Lean v4.33.1)

analysisoptimizationstochastic-networks

The weighted α\alphaα-fair objective is defined by two formulas, one for α≠1\alpha \ne 1α=1 and one for α=1\alpha = 1α=1:

G(X)=∑rwrnrαXr1−α1−α,G(X)=∑rwrnrlog⁡Xr.G(X) = \sum_r w_r n_r^{\alpha}\frac{X_r^{1-\alpha}}{1-\alpha}, \qquad\qquad G(X) = \sum_r w_r n_r \log X_r .G(X)=r∑​wr​nrα​1−αXr1−α​​,G(X)=r∑​wr​nr​logXr​.

Exercise 8.3 observes that this is a single family, because the derivative has the same form in both cases. For α∈(0,∞)\alpha \in (0,\infty)α∈(0,∞), positive weights, positive flow counts and a positive rate vector,

∂G∂Xr(X)=wr nrα Xr−α.\frac{\partial G}{\partial X_r}(X) = w_r\,n_r^{\alpha}\,X_r^{-\alpha}.∂Xr​∂G​(X)=wr​nrα​Xr−α​.

This is why α=1\alpha = 1α=1 is the natural continuation of the family rather than a separate case stapled on: the singularity in Xr1−α/(1−α)X_r^{1-\alpha}/(1-\alpha)Xr1−α​/(1−α) at α=1\alpha = 1α=1 is removable once a constant is absorbed, and the derivative never notices it. It is also the formula the stability argument uses, through the tangent-plane inequality, where the same expression appears for every α\alphaα at once.

Formalization Note The partial derivative is stated as an ordinary derivative in the rrr-th coordinate with the others held fixed, so no partial-derivative API is presupposed. The claim covers the defined-by-cases objective, so proving it means handling both branches and checking they agree.

Preamble
import Mathlib
import Definitions.Def_KellyStochasticNetworks_Wardrop
import Definitions.Def_KellyStochasticNetworks_Congestion
import Definitions.Def_KellyStochasticNetworks_FlowLevel
Formal statement
namespace KellyStochasticNetworks

theorem alpha_fair_deriv {R : ℕ} (w n : Fin R → ℝ) (α : ℝ) (hα : 0 < α)
    (hw : ∀ r, 0 < w r) (hn : ∀ r, 0 < n r) (X : Fin R → ℝ) (hX : ∀ r, 0 < X r) (r : Fin R) :
    HasDerivAt (fun t : ℝ => alphaFairObjective w n α (Function.update X r t))
      (w r * n r ^ α * X r ^ (-α)) (X r) := by sorry

end KellyStochasticNetworks
Source
Kelly & Yudovina, Stochastic Networks, CUP 2014, p. 189 (PDF p. 197), Exercise 8.3: 'Check that the objective function of the problem (8.1) is concave for each of the three cases 0 < alpha < 1, alpha = 1 and 1 < alpha < infinity. Show that the derivative of the objective function with respect to x_r has the same form, namely w_r n_r x_r^{-alpha}, for all alpha in (0, infinity).' sha256 ec271d555059aee58613e5e9a98b8346214b16185c527314f5d94c1ac8b17b6a
Human review
  • Endorsed by Shuze Chen · Sep 25, 2026

    Confirmed by the moderator at approval.

  • Endorsed by naimengye · Sep 25, 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