InformationTheory

InformationTheory.Shannon.BroadcastChannel.Superposition.MoreCapable

source

Broadcast channel — the capacity region of a more capable channel #

IsBCMoreCapable compares the two marginal channels under every input law. The converse for a more capable channel needs that comparison in two further shapes: conditioned on an auxiliary variable, and read on a five-tuple law of the UV outer region. Both are averages of the unconditional statement over the conditional input laws, which is exactly what the mixture identity for conditional mutual information delivers, so the comparison transports with no new analysis.

What the comparison buys is the sum-rate inequality I(V; Y₁) + I(X; Y₂ ∣ V) ≤ I(X; Y₁), the fourth constraint of the UV outer region, which the less noisy converse discards and the more capable one needs. The companion bound at the receiver-2 corner, I(U; Y₂) ≤ I(X; Y₁), is the same comparison composed with the data processing inequality along U → X → Y₂; it is what the sum constraint degenerates to once the first rate is clamped at zero.

The right-hand side of both bounds is uvInfoJoint, the information the input carries about the first output. It is the slot of a five-tuple law that the two-constraint inner bound never needed, and this file records the fact the full-support step asks of it beyond its invariance under relabeling: it is concave in the law, so a positive multiple of it survives the mixture that repairs full support.

Keeping that slot is what lets the inner bound keep the sum-rate constraint a general broadcast channel needs, instead of the two-constraint region that is exact only over a less noisy one. Time sharing between a rate pair and the corner where the cloud auxiliary is constant meets the two separate constraints at once, and the sum constraint survives the segment because the input-output slot is the same at both of its endpoints; perturbing the achievability pair the segment lands on toward the uniform law then repairs its support, at a cost one weight covers for all three slots at once. Truncating the auxiliary of a law of the outer region moves that law onto a finite alphabet, and leaves the input-output slot alone as well, so both costs vanish along a single sequence of indices and the inner bound, being a closure, recovers the rate pair itself. Combined with the outer bound and with the achievability of the inner one, this describes the capacity region of a more capable broadcast channel by a single-letter expression. It subsumes the description of a less noisy one, IsBCLessNoisy being the stronger comparison.

The sum constraint is carried with the first rate clamped at zero, max R₁ 0 + R₂, which is the form the achievability theorem takes and the form under which the inner bound is achievable with no comparison between the two receivers. A negative first rate is then met at the receiver-2 corner rather than by the sum-rate slot of the outer region, which is where the second of the two comparison bounds is spent.

Main statements #

Implementation notes #

Transporting the comparison means exchanging the measure inside an information, which rw cannot do: condMutualInfo takes the finiteness instance on its measure argument, so rewriting that argument leaves a motive that is not type correct. condMutualInfo_congr_measure performs the exchange by substitution instead, and mutualInfo_congr_pair is the coarser form, where the two ambient measures may differ as long as the joint law of the compared pair agrees. mutualInfo_compProd_out₁ and mutualInfo_compProd_out₂ read a marginal channel off a composition product, which is what puts the comparison in the mutualInfoOfChannel form IsBCMoreCapable is stated in.

The two marginal channels under a single input law #

theorem

InformationTheory.Shannon.BroadcastChannel.mutualInfo_compProd_out₁

source
{α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] (p : MeasureTheory.Measure α) [MeasureTheory.IsProbabilityMeasure p] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] :
Used by
    theorem

    InformationTheory.Shannon.BroadcastChannel.mutualInfo_compProd_out₂

    source
    {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] (p : MeasureTheory.Measure α) [MeasureTheory.IsProbabilityMeasure p] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] :
    Used by

      The conditional form of the more capable comparison #

      theorem

      InformationTheory.Shannon.BroadcastChannel.bcJointDistribution_eq_compProd

      source
      Used by
        theorem

        InformationTheory.Shannon.BroadcastChannel.compProd_comap_snd_apply

        source
        {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} [MeasurableSpace U] (K : ProbabilityTheory.Kernel U α) [ProbabilityTheory.IsMarkovKernel K] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (u : U) :
        Used by
          theorem

          InformationTheory.Shannon.BroadcastChannel.condMutualInfo_bcJointDistribution_out₁_eq_lintegral

          source
          {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} [MeasurableSpace U] [StandardBorelSpace α] [Nonempty α] [Countable U] [MeasurableSingletonClass U] [StandardBorelSpace β₁] [Nonempty β₁] (pU : MeasureTheory.Measure U) [MeasureTheory.IsProbabilityMeasure pU] (K : ProbabilityTheory.Kernel U α) [ProbabilityTheory.IsMarkovKernel K] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] :
          (condMutualInfo (bcJointDistribution pU K W) (fun (q : U × α × β₁ × β₂) => q.2.1) (fun (q : U × α × β₁ × β₂) => q.2.2.1) fun (q : U × α × β₁ × β₂) => q.1) = ∫⁻ (u : U), ChannelCoding.mutualInfoOfChannel (K u) (ProbabilityTheory.Kernel.fst W) pU
          Used by
            theorem

            InformationTheory.Shannon.BroadcastChannel.condMutualInfo_bcJointDistribution_out₂_eq_lintegral

            source
            {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} [MeasurableSpace U] [StandardBorelSpace α] [Nonempty α] [Countable U] [MeasurableSingletonClass U] [StandardBorelSpace β₂] [Nonempty β₂] (pU : MeasureTheory.Measure U) [MeasureTheory.IsProbabilityMeasure pU] (K : ProbabilityTheory.Kernel U α) [ProbabilityTheory.IsMarkovKernel K] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] :
            (condMutualInfo (bcJointDistribution pU K W) (fun (q : U × α × β₁ × β₂) => q.2.1) (fun (q : U × α × β₁ × β₂) => q.2.2.2) fun (q : U × α × β₁ × β₂) => q.1) = ∫⁻ (u : U), ChannelCoding.mutualInfoOfChannel (K u) (ProbabilityTheory.Kernel.snd W) pU
            Used by
              theorem

              InformationTheory.Shannon.BroadcastChannel.IsBCMoreCapable.condMutualInfo_le

              source
              {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} [MeasurableSpace U] [StandardBorelSpace α] [Nonempty α] [Countable U] [MeasurableSingletonClass U] [StandardBorelSpace β₁] [Nonempty β₁] [StandardBorelSpace β₂] [Nonempty β₂] {W : BCChannel α β₁ β₂} [ProbabilityTheory.IsMarkovKernel W] (hmc : IsBCMoreCapable W) (pU : MeasureTheory.Measure U) [MeasureTheory.IsProbabilityMeasure pU] (K : ProbabilityTheory.Kernel U α) [ProbabilityTheory.IsMarkovKernel K] :
              (condMutualInfo (bcJointDistribution pU K W) (fun (q : U × α × β₁ × β₂) => q.2.1) (fun (q : U × α × β₁ × β₂) => q.2.2.2) fun (q : U × α × β₁ × β₂) => q.1) condMutualInfo (bcJointDistribution pU K W) (fun (q : U × α × β₁ × β₂) => q.2.1) (fun (q : U × α × β₁ × β₂) => q.2.2.1) fun (q : U × α × β₁ × β₂) => q.1
              Used by

                The Markov chain reaching the second receiver #

                theorem

                InformationTheory.Shannon.BroadcastChannel.IsUVChannelLaw.isMarkovChain_U_X_Y₂

                source
                {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [StandardBorelSpace α] [Nonempty α] [MeasurableSpace β₁] [StandardBorelSpace β₁] [Nonempty β₁] [MeasurableSpace β₂] [StandardBorelSpace β₂] [Nonempty β₂] {U : Type u_4} {V : Type u_5} [MeasurableSpace U] [StandardBorelSpace U] [Nonempty U] [MeasurableSpace V] [StandardBorelSpace V] [Nonempty V] {W : BCChannel α β₁ β₂} [ProbabilityTheory.IsMarkovKernel W] {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) :
                IsMarkovChain ν (fun (q : U × V × α × β₁ × β₂) => q.1) (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.2.2.2.2
                Used by

                  The more capable comparison on a channel law #

                  theorem

                  InformationTheory.Shannon.BroadcastChannel.uvInfoJoint_eq_uvInfo₁_add_condMutualInfo

                  source
                  {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [StandardBorelSpace α] [Nonempty α] [MeasurableSpace β₁] [StandardBorelSpace β₁] [Nonempty β₁] [MeasurableSpace β₂] [StandardBorelSpace β₂] [Nonempty β₂] {U : Type u_4} [MeasurableSpace U] {V : Type u_5} [MeasurableSpace V] [StandardBorelSpace U] [Nonempty U] [StandardBorelSpace V] [Nonempty V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) :
                  uvInfoJoint ν = uvInfo₁ ν + condMutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) (fun (q : U × V × α × β₁ × β₂) => q.2.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.2.1
                  Used by
                    theorem

                    InformationTheory.Shannon.BroadcastChannel.condMutualInfo_out₂_le_out₁_of_moreCapable

                    source
                    {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [StandardBorelSpace α] [Nonempty α] [MeasurableSpace β₁] [StandardBorelSpace β₁] [Nonempty β₁] [MeasurableSpace β₂] [StandardBorelSpace β₂] [Nonempty β₂] {U : Type u_4} [MeasurableSpace U] {V : Type u_5} [MeasurableSpace V] [Countable V] [MeasurableSingletonClass V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hmc : IsBCMoreCapable W) {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) :
                    (condMutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) (fun (q : U × V × α × β₁ × β₂) => q.2.2.2.2) fun (q : U × V × α × β₁ × β₂) => q.2.1) condMutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) (fun (q : U × V × α × β₁ × β₂) => q.2.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.2.1
                    Used by
                      theorem

                      InformationTheory.Shannon.BroadcastChannel.uvInfoSum₁_le_uvInfoJoint_of_moreCapable

                      source
                      Used by
                        theorem

                        InformationTheory.Shannon.BroadcastChannel.mutualInfo_out₂_le_out₁_of_moreCapable

                        source
                        {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} {V : Type u_5} [MeasurableSpace U] [MeasurableSpace V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hmc : IsBCMoreCapable W) {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) :
                        (mutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.2.2.2.2) mutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.2.2.2.1
                        Used by
                          theorem

                          InformationTheory.Shannon.BroadcastChannel.uvInfo₂_le_uvInfoJoint_of_moreCapable

                          source
                          {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] {U : Type u_4} {V : Type u_5} [MeasurableSpace U] [MeasurableSpace V] [StandardBorelSpace α] [Nonempty α] [StandardBorelSpace β₁] [Nonempty β₁] [StandardBorelSpace β₂] [Nonempty β₂] [StandardBorelSpace U] [Nonempty U] [StandardBorelSpace V] [Nonempty V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hmc : IsBCMoreCapable W) {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) :
                          Used by

                            The input-output slot under mixing #

                            theorem

                            InformationTheory.Shannon.BroadcastChannel.mul_uvInfoJoint_le_uvInfoJoint_uvMixLaw

                            source
                            {α : Type u_1} {β₁ : Type u_2} {β₂ : Type u_3} [MeasurableSpace α] [StandardBorelSpace α] [Nonempty α] [MeasurableSpace β₁] [StandardBorelSpace β₁] [Nonempty β₁] [MeasurableSpace β₂] [StandardBorelSpace β₂] [Nonempty β₂] {U : Type u_4} {V : Type u_5} [MeasurableSpace U] [StandardBorelSpace U] [Nonempty U] [MeasurableSpace V] [StandardBorelSpace V] [Nonempty V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] {ν σ : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] [MeasureTheory.IsProbabilityMeasure σ] ( : IsUVChannelLaw W ν) ( : IsUVChannelLaw W σ) (lam : ENNReal) (hlam : lam 1) :
                            lam * uvInfoJoint ν uvInfoJoint (uvMixLaw ν σ lam)
                            Used by
                              theorem

                              InformationTheory.Shannon.BroadcastChannel.mul_uvInfoJoint_le_uvInfoJoint_uvPerturbLaw

                              source
                              {α : Type u_1} [Fintype α] [Nonempty α] [MeasurableSpace α] [StandardBorelSpace α] {β₁ : Type u_2} [MeasurableSpace β₁] [StandardBorelSpace β₁] [Nonempty β₁] {β₂ : Type u_3} [MeasurableSpace β₂] [StandardBorelSpace β₂] [Nonempty β₂] {U : Type u_4} [Fintype U] [Nonempty U] [MeasurableSpace U] [StandardBorelSpace U] {V : Type u_5} [MeasurableSpace V] [StandardBorelSpace V] [Nonempty V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) (v₀ : V) {lam : ENNReal} (hlam : lam 1) :
                              lam * uvInfoJoint ν uvInfoJoint (uvPerturbLaw W ν v₀ lam)
                              Used by

                                Repairing the support with the sum constraint #

                                theorem

                                InformationTheory.Shannon.BroadcastChannel.exists_fullSupport_bcInfo_ge_sumRate_of_isUVChannelLaw

                                source
                                {α : Type u} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] {β₁ : Type u_1} [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] {β₂ : Type u_2} [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] {V : Type u_3} [MeasurableSpace V] [StandardBorelSpace V] [Nonempty V] {U : Type u} [Fintype U] [Nonempty U] [MeasurableSpace U] [MeasurableSingletonClass U] [StandardBorelSpace U] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] {ν : MeasureTheory.Measure (U × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) {R₁ R₂ δ : } ( : 0 < δ) (h₁ : R₁ (condMutualInfo ν (fun (q : U × V × α × β₁ × β₂) => q.2.2.1) (fun (q : U × V × α × β₁ × β₂) => q.2.2.2.1) fun (q : U × V × α × β₁ × β₂) => q.1).toReal) (h₂ : R₂ (uvInfo₂ ν).toReal) (h₃ : max R₁ 0 + R₂ (uvInfoJoint ν).toReal) :
                                ∃ (pU : MeasureTheory.Measure U) (_ : MeasureTheory.IsProbabilityMeasure pU) (_ : ∀ (u : U), 0 < pU.real {u}) (K : ProbabilityTheory.Kernel U α) (_ : ProbabilityTheory.IsMarkovKernel K) (_ : ∀ (u : U) (a : α), 0 < (K u).real {a}), R₁ - δ bcInfo₁ pU K W R₂ - δ bcInfo₂ pU K W max (R₁ - δ) 0 + (R₂ - δ) bcInfoJoint pU K W
                                Used by
                                  theorem

                                  InformationTheory.Shannon.BroadcastChannel.exists_fullSupport_bcInfo_ge_sumRate

                                  source
                                  {α : Type u} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] {β₁ : Type u_1} [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] {β₂ : Type u_2} [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] {U : Type u} [Fintype U] [Nonempty U] [MeasurableSpace U] [MeasurableSingletonClass U] [StandardBorelSpace U] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (pU : MeasureTheory.Measure U) [MeasureTheory.IsProbabilityMeasure pU] (K : ProbabilityTheory.Kernel U α) [ProbabilityTheory.IsMarkovKernel K] {R₁ R₂ δ : } ( : 0 < δ) (h₁ : R₁ bcInfo₁ pU K W) (h₂ : R₂ bcInfo₂ pU K W) (h₃ : max R₁ 0 + R₂ bcInfoJoint pU K W) :
                                  ∃ (pU' : MeasureTheory.Measure U) (_ : MeasureTheory.IsProbabilityMeasure pU') (_ : ∀ (u : U), 0 < pU'.real {u}) (K' : ProbabilityTheory.Kernel U α) (_ : ProbabilityTheory.IsMarkovKernel K') (_ : ∀ (u : U) (a : α), 0 < (K' u).real {a}), R₁ - δ bcInfo₁ pU' K' W R₂ - δ bcInfo₂ pU' K' W max (R₁ - δ) 0 + (R₂ - δ) bcInfoJoint pU' K' W
                                  Used by
                                    theorem

                                    InformationTheory.Shannon.BroadcastChannel.sub_mem_bcSuperpositionRegionSumRate_of_isUVChannelLaw

                                    source
                                    {α : Type u} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] {β₁ : Type u_1} [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] {β₂ : Type u_2} [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] {V : Type u_3} [MeasurableSpace V] [StandardBorelSpace V] [Nonempty V] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] {m : } {ν : MeasureTheory.Measure (Marton.bcAuxAlphabet m × V × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) {R₁ R₂ δ : } ( : 0 < δ) (h₁ : R₁ (uvInfo₁ ν).toReal) (h₂ : R₂ (uvInfo₂ ν).toReal) (hs₂ : R₁ + R₂ (uvInfoSum₂ ν).toReal) (hs₁ : max R₁ 0 + R₂ (uvInfoJoint ν).toReal) :
                                    Used by

                                      The reverse inclusion #

                                      theorem

                                      InformationTheory.Shannon.BroadcastChannel.uvInfoJoint_uvQuantizeLaw

                                      source
                                      {α : Type u} {β₁ : Type u_1} {β₂ : Type u_2} [MeasurableSpace α] [MeasurableSpace β₁] [MeasurableSpace β₂] (ν : MeasureTheory.Measure ( × × α × β₁ × β₂)) (m : ) :
                                      Used by
                                        theorem

                                        InformationTheory.Shannon.BroadcastChannel.sub_mem_bcSuperpositionRegionSumRate_of_mem_uvRegion

                                        source
                                        {α : Type u} {β₁ : Type u_1} {β₂ : Type u_2} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hmc : IsBCMoreCapable W) {ν : MeasureTheory.Measure ( × × α × β₁ × β₂)} [MeasureTheory.IsProbabilityMeasure ν] (h : IsUVChannelLaw W ν) {p : × } (hp : p uvRegion ν) (m : ) {δ : } ( : 0 < δ) :
                                        Used by
                                          theorem

                                          InformationTheory.Shannon.BroadcastChannel.mem_bcSuperpositionRegionSumRate_of_mem_uvRegion

                                          source
                                          Used by
                                            theorem

                                            InformationTheory.Shannon.BroadcastChannel.bc_moreCapable_uv_subset_superposition

                                            source

                                            The UV outer region of a more capable broadcast channel is contained in the three-constraint superposition inner bound over the full-support achievability pairs. The channel needs no support hypothesis here: the inclusion compares two single-letter regions, and positive mass on every output pair is asked for only where the inner bound is turned into codes (bcSuperpositionRegionSumRate_subset_capacity).

                                            Used by

                                              The capacity region of a more capable channel #

                                              theorem

                                              InformationTheory.Shannon.BroadcastChannel.bc_moreCapable_capacity_eq_uv

                                              source
                                              {α : Type u} {β₁ : Type u_1} {β₂ : Type u_2} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hW : ∀ (a : α) (b : β₁ × β₂), 0 < (W a).real {b}) (hmc : IsBCMoreCapable W) :

                                              The capacity region of a more capable broadcast channel whose transition law gives every output pair positive mass is its UV outer region bcOuterRegionUV, a single-letter expression in the four information slots of a five-tuple law.

                                              Used by
                                                theorem

                                                InformationTheory.Shannon.BroadcastChannel.bc_degraded_capacity_eq_uv

                                                source
                                                {α : Type u} {β₁ : Type u_1} {β₂ : Type u_2} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hW : ∀ (a : α) (b : β₁ × β₂), 0 < (W a).real {b}) (hdeg : IsBCDegraded W) :

                                                The capacity region of a physically degraded broadcast channel whose transition law gives every output pair positive mass is its UV outer region. Degradedness is the strongest of the three comparisons, so the single-letter characterization of a more capable channel applies with no further analysis.

                                                Used by
                                                  theorem

                                                  InformationTheory.Shannon.BroadcastChannel.bc_moreCapable_superposition_eq_capacity

                                                  source
                                                  {α : Type u} {β₁ : Type u_1} {β₂ : Type u_2} [Fintype α] [Nonempty α] [MeasurableSpace α] [MeasurableSingletonClass α] [StandardBorelSpace α] [Fintype β₁] [Nonempty β₁] [MeasurableSpace β₁] [MeasurableSingletonClass β₁] [StandardBorelSpace β₁] [Fintype β₂] [Nonempty β₂] [MeasurableSpace β₂] [MeasurableSingletonClass β₂] [StandardBorelSpace β₂] (W : BCChannel α β₁ β₂) [ProbabilityTheory.IsMarkovKernel W] (hW : ∀ (a : α) (b : β₁ × β₂), 0 < (W a).real {b}) (hmc : IsBCMoreCapable W) :
                                                  Used by