Documentation

IntroductoryLecturesOnConvexOptimization_Nesterov_2004.Chap05.Definition_5_0_21

noncomputable def selfConcordantOmega :
(Set.Ioi (-1))

Definition 5.0.21 (1): the auxiliary function ω(t) = t - log (1 + t) on (-1, ∞).

Instances For
    noncomputable def selfConcordantOmegaStar :
    (Set.Iio 1)

    Definition 5.0.21 (2): the auxiliary function ω_*(τ) = -τ - log (1 - τ) on (-∞, 1).

    Instances For
      noncomputable def selfConcordantOmegaDeriv :
      (Set.Ioi (-1))

      The derivative branch ω'(t) = t / (1 + t) of ω, defined on its natural domain (-1, ∞).

      Instances For
        noncomputable def selfConcordantOmegaPrimeStar :
        (Set.Iio 1)

        The inverse branch ω'_*(τ) = τ / (1 - τ) associated with ω and ω_*, defined on its natural domain (-∞, 1).

        Instances For
          Instances For
            Instances For
              Instances For
                Instances For
                  theorem mf_mul_lt_one_of_lt_inv {Mf : NNReal} {r : } (hr : r < 1 / Mf) :
                  Mf * r < 1

                  If r < 1 / M_f, then the scaled quantity M_f r lies in the natural domain of ω_*.

                  theorem neg_one_lt_mf_mul_of_nonneg {Mf : NNReal} {r : } (hr : 0 r) :
                  -1 < Mf * r

                  If r ≥ 0, then the scaled quantity M_f r lies in the natural domain of ω.

                  def selfConcordantOmegaArg (Mf : NNReal) (r : ) (hr : -1 < Mf * r) :
                  (Set.Ioi (-1))

                  The canonical ω argument attached to a scalar r whose scaled value lies in the natural domain of ω.

                  Instances For
                    @[simp]
                    theorem coe_selfConcordantOmegaArg (Mf : NNReal) (r : ) (hr : -1 < Mf * r) :
                    (selfConcordantOmegaArg Mf r hr) = Mf * r

                    Coercing selfConcordantOmegaArg Mf r hr back to recovers M_f r.

                    noncomputable def selfConcordantOmegaOneHalfArg :
                    (Set.Ioi (-1))

                    The textbook point 1 / 2 in the natural domain (-1, ∞) of ω.

                    Instances For
                      @[simp]

                      Coercing selfConcordantOmegaOneHalfArg back to recovers 1 / 2.

                      noncomputable def selfConcordantOmegaAtOneHalf :

                      The Chapter 5 textbook constant ω(1 / 2).

                      Instances For
                        def selfConcordantOmegaStarArg (Mf : NNReal) (r : ) (hr : Mf * r < 1) :
                        (Set.Iio 1)

                        The canonical ω_* argument attached to a scalar r satisfying (Mf : ℝ) * r < 1.

                        Instances For
                          @[simp]
                          theorem coe_selfConcordantOmegaStarArg (Mf : NNReal) (r : ) (hr : Mf * r < 1) :
                          (selfConcordantOmegaStarArg Mf r hr) = Mf * r

                          Coercing selfConcordantOmegaStarArg Mf r hr back to recovers M_f r.

                          @[simp]
                          theorem selfConcordantOmega_apply (t : (Set.Ioi (-1))) :
                          selfConcordantOmega t = t - Real.log (1 + t)

                          Evaluating ω at a point of (-1, ∞) recovers the explicit formula t - log (1 + t).

                          @[simp]
                          theorem selfConcordantOmegaStar_apply (τ : (Set.Iio 1)) :
                          selfConcordantOmegaStar τ = -τ - Real.log (1 - τ)

                          Evaluating ω_* at a point of (-∞, 1) recovers the textbook formula -τ - log (1 - τ).

                          @[simp]
                          theorem selfConcordantOmegaDeriv_apply (t : (Set.Ioi (-1))) :
                          selfConcordantOmegaDeriv t = t / (1 + t)

                          Evaluating ω' at a point of (-1, ∞) recovers the explicit derivative formula t / (1 + t).

                          @[simp]
                          theorem selfConcordantOmegaPrimeStar_apply (τ : (Set.Iio 1)) :
                          selfConcordantOmegaPrimeStar τ = τ / (1 - τ)

                          Evaluating ω'_* at a point of (-∞, 1) recovers the explicit inverse-branch formula τ / (1 - τ).