AFTD Auto-Formalizing Theoretical Domains

potential_pos

theoremverifiedimported Lean

Positivity of the Hedge potential. Fix a learning rate $\eta \in \bbr$, a loss sequence $\ell$ for $N \ge 1$ experts over $T$ rounds, and a time $t$. Then the potential at time $t$ is strictly positive: \[ W_t \;=\; \sum_{i=1}^{N} \exp\!\bigl(-\eta \cdot L_t(i)\bigr) \;>\; 0, \] where $L_t(i) = \sum_{s < t} \ell_s(i)$ is the cumulative loss of expert $i$ through the first $t$ rounds.

Statement

set_option relaxedAutoImplicit false in
set_option autoImplicit false in
open Real Finset BigOperators in
lemma potential_pos {N T : ℕ} [NeZero N] (η : ℝ) (ℓ : LossSeq N T) (t : ℕ) :
    0 < potential η ℓ t
Source
TCSlib, `potential_pos`
Lean proof
By Karim Abdel Sadek, Mark Bedaywi, from github.com/Shilun-Allan-Li/tcslib at c8a02591b1e5 (Copyright (c) 2026 Karim Abdel Sadek and Mark Bedaywi. All rights reserved, Apache-2.0). Compiled here against leanprover/lean4:v4.33.1, Mathlib v4.33.1; written for leanprover/lean4:v4.25.0.
Verified
07 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 3
  • hedgeWeight_posPositivity of Hedge weights. Fix a learning rate $\eta \in \bbr$, a loss sequence $\ell$ for $N$ experts over $T$ rounds, a time $t$, and…
  • potentialThe potential (sum of unnormalized weights) at time $t$ is \[ W_t \;=\; \sum_{i=1}^{N} w_t(i). \] The potential telescopes to…
  • LossSeqA loss sequence $\ell$ for $N$ experts over $T$ rounds is a function $\ell : \mathrm{Fin}\,T \to \mathrm{Fin}\,N \to \mathbb{R}$,…
Used by 5
  • log_potential_step_tightTight per-step logarithmic potential bound for Hedge. Fix $N \ge 1$ experts and a learning rate $\eta > 0$, and let $\ell$ be a valid loss…
  • potential_ratio_lePotential ratio bound for Hedge. Fix $N \ge 1$ experts, a horizon of $T$ rounds, a learning rate $\eta > 0$, and a valid loss sequence…
  • log_potential_stepOne-step logarithmic potential bound for Hedge. Fix $N \ge 1$ experts and $T$ rounds, a learning rate $\eta > 0$, and a valid loss…
  • hedgeDist_nonnegNonnegativity of the Hedge distribution. Fix $N \ge 1$ experts and $T$ rounds, a learning rate $\eta \in \bbr$, and a loss sequence…
  • hedgeDist_sumThe Hedge distribution sums to one. Fix a number of experts $N \ge 1$, a horizon $T$, a learning rate $\eta \in \bbr$, a loss sequence…

Lean source view module on GitHub

set_option relaxedAutoImplicit false inset_option autoImplicit false inopen Real Finset BigOperators in/-- The potential is strictly positive: it is a nonempty sum of positive weights. -/lemma potential_pos {N T : ℕ} [NeZero N] (η : ℝ) (ℓ : LossSeq N T) (t : ℕ) :    0 < potential η ℓ t := by  apply Finset.sum_pos  · intro i _    exact hedgeWeight_pos η ℓ t i  · exact Finset.univ_nonempty
See it in the graphDiscuss this resultChallenge it