Physlib

PhyslibAlpha.ProbabilisticTheory.Measurement.Finite

Measurements with finitely many outcomes

i. Overview

On a finite discrete outcome space, every observable of the outcome is a combination of the indicators of single outcomes. A measurement is therefore determined by the effects of its single outcomes, and these add up to `1`. Conversely, a family of effects adding up to `1` is a measurement: send the observable `f` to `∑ x, f x • a x`. Normality is automatic, since an increasing sequence of functions on finitely many points converges uniformly; this uses that the system is Archimedean.

ii. Key results

- `BoundedMeasurable.eq_sum_indicator` : an observable of a finite classical system is a combination of point indicators. - `UnitalPositiveLinearMap.isNormal_of_fintype` : a channel from a finite classical system into an Archimedean system is normal. - `Measurement.ofAtoms` : the measurement of a family of effects adding up to `1`. - `Measurement.eq_ofAtoms` : a measurement with finitely many outcomes is determined by the effects of its outcomes.

iii. Table of contents

  • A. Observables of a finite classical system
  • B. Normality is automatic
  • C. Measurements from their outcome effects

A. Observables of a finite classical system

B. Normality is automatic

C. Measurements from their outcome effects

10 declarations

theorem

∑x1{x}=1\sum_x \mathbb{1}_{\{x\}} = 1 in Bounded Measurable Functions

In the space of bounded measurable functions on a finite outcome space Ω\Omega, the sum of the indicator functions 1{x}\mathbb{1}_{\{x\}} for all individual outcomes x∈Ωx \in \Omega is equal to the unit function 11.

theorem

For a monotone sequence fn↑gf_n \uparrow g on a finite set, ∃n,g≤fn+ε1\exists n, g \le f_n + \varepsilon \mathbf{1}

Let Ω\Omega be a finite sample space and BoundedMeasurable(Ω)\text{BoundedMeasurable}(\Omega) be the space of bounded measurable functions on Ω\Omega. Suppose f:N→BoundedMeasurable(Ω)f : \mathbb{N} \to \text{BoundedMeasurable}(\Omega) is a monotone increasing sequence of functions and g∈BoundedMeasurable(Ω)g \in \text{BoundedMeasurable}(\Omega) is the least upper bound (supremum) of the range of ff. Then for any real number ε>0\varepsilon > 0, there exists a natural number nn such that g≤fn+ε1g \le f_n + \varepsilon \mathbf{1}, where 1\mathbf{1} denotes the constant unit function.

theorem

Channels from finite classical systems are normal

Let Ω\Omega be a finite sample space. Let BoundedMeasurable(Ω)\text{BoundedMeasurable}(\Omega) be the space of bounded measurable functions on Ω\Omega, and let EE be an Archimedean order unit space. Any channel Φ:BoundedMeasurable(Ω)→E\Phi : \text{BoundedMeasurable}(\Omega) \to E (that is, a unital positive linear map) is normal, meaning that it preserves the suprema of directed sets.

definition

Quantum channel defined by effects {ax}\{a_x\} with ∑ax=1\sum a_x = 1

For a finite outcome space Ω\Omega and an order unit space EE, given a family of effects {ax}x∈Ω\{a_x\}_{x \in \Omega} in EE (where each 0≤ax≤10 \le a_x \le 1) such that they satisfy the normalization condition ∑x∈Ωax=1\sum_{x \in \Omega} a_x = 1, this definition constructs a quantum channel (a unital positive linear map) Φ:BoundedMeasurable(Ω)→E\Phi: \text{BoundedMeasurable}(\Omega) \to E. For any observable f∈BoundedMeasurable(Ω)f \in \text{BoundedMeasurable}(\Omega), the channel is defined by the mapping: Φ(f)=∑x∈Ωf(x)⋅ax\Phi(f) = \sum_{x \in \Omega} f(x) \cdot a_x where f(x)f(x) are the values of the observable at each outcome xx.

definition

Measurement defined by effects {ax}\{a_x\} with ∑ax=1\sum a_x = 1

For a finite outcome space Ω\Omega and an order unit space EE, let {ax}x∈Ω\{a_x\}_{x \in \Omega} be a family of effects in EE (where 0≤ax≤10 \le a_x \le 1) such that they satisfy the normalization condition ∑x∈Ωax=1\sum_{x \in \Omega} a_x = 1. This definition constructs a measurement MM whose underlying channel Φ:BoundedMeasurable(Ω)→E\Phi: \text{BoundedMeasurable}(\Omega) \to E maps an observable ff to the weighted sum of effects: Φ(f)=∑x∈Ωf(x)ax\Phi(f) = \sum_{x \in \Omega} f(x) a_x In this measurement, each outcome x∈Ωx \in \Omega corresponds to the effect axa_x.

theorem

The channel of a measurement defined by atoms {ax}\{a_x\} is Φ(f)=∑xf(x)ax\Phi(f) = \sum_x f(x) a_x

Let Ω\Omega be a finite outcome space and EE be an order unit space. Suppose {ax}x∈Ω\{a_x\}_{x \in \Omega} is a family of effects in EE (where 0≤ax≤10 \le a_x \le 1) that satisfy the normalization condition ∑x∈Ωax=1\sum_{x \in \Omega} a_x = 1. For the measurement MM constructed from these effects, the associated quantum channel Φ:BoundedMeasurable(Ω)→E\Phi: \text{BoundedMeasurable}(\Omega) \to E maps any observable ff to the weighted sum of effects: Φ(f)=∑x∈Ωf(x)ax\Phi(f) = \sum_{x \in \Omega} f(x) a_x where f(x)f(x) are the values of the observable at each outcome xx.

theorem

M(s)=∑x∈saxM(s) = \sum_{x \in s} a_x for a measurement MM defined by atomic effects {ax}\{a_x\}

Let Ω\Omega be a finite outcome space and EE be an order unit space. Given a family of effects {ax}x∈Ω\{a_x\}_{x \in \Omega} in EE satisfying the normalization condition ∑x∈Ωax=1\sum_{x \in \Omega} a_x = 1, let MM be the measurement constructed from these atomic effects. For any measurable subset s⊆Ωs \subseteq \Omega, the value of the measurement MM evaluated on ss is the sum of the effects of the outcomes contained in ss: M(s)=∑x∈sax M(s) = \sum_{x \in s} a_x

theorem

The measurement MM defined by atoms {ax}\{a_x\} satisfies M({x})=axM(\{x\}) = a_x

Let Ω\Omega be a finite sample space and EE be an order unit space. Suppose we have a family of effects {ax}x∈Ω\{a_x\}_{x \in \Omega} in EE that satisfy the normalization condition ∑x∈Ωax=1\sum_{x \in \Omega} a_x = 1. Let MM be the measurement constructed from these effects (defined via `ofAtoms`). For any outcome x∈Ωx \in \Omega such that the singleton set {x}\{x\} is measurable, the effect assigned by the measurement MM to the outcome xx is exactly axa_x: M({x})=ax M(\{x\}) = a_x

theorem

The sum of outcome effects ∑x∈ΩM({x})\sum_{x \in \Omega} M(\{x\}) equals 11 for a measurement MM

Let Ω\Omega be a finite sample space and MM be a measurement on Ω\Omega with values in an order unit space EE. For each outcome x∈Ωx \in \Omega, let M({x})M(\{x\}) denote the effect associated with the singleton set {x}\{x\}. Then the sum of these effects over all possible outcomes in Ω\Omega is equal to the unit element 11 of EE: ∑x∈ΩM({x})=1\sum_{x \in \Omega} M(\{x\}) = 1

theorem

Measurements with finite outcomes are determined by their singleton effects

Let Ω\Omega be a finite outcome space and EE an order unit space. Suppose MM and NN are two measurements from Ω\Omega to EE. If for every individual outcome x∈Ωx \in \Omega, the effect assigned by MM to the singleton set {x}\{x\} is equal to the effect assigned by NN to {x}\{x\}, then the measurements MM and NN are equal.