Physlib

PhyslibAlpha.AlgebraicFramework.Measurement.Basic

Finite-outcome measurements

A finite-outcome measurement is a finite family of effects whose total is the order unit. The outcome *type* `ι` need not itself be finite — only finitely many outcomes need actually occur, recorded by an explicit `Finset ι` of outcomes — so a measurement can, for instance, be labeled by all of `ℕ` or `ℝ` while only ever registering finitely many of those labels. Evaluating a measurement in a state gives a probability distribution on its (finite) outcome set, directly in `ℝ`: a state is already a genuine linear functional, so its values on the (bounded) effects of a measurement are already finite nonnegative reals summing to `1`, with no detour through `Weight`'s `[0, ∞]`-valued arithmetic needed.

This sits outside `OrderUnit/`: it is a derived notion built *from* `Effect` and `State`, not part of the order-unit algebra itself, the same way `Traciality.lean` sits outside `OrderUnit/` despite being built from `Weight`. `FiniteOutcome.lean` gives the equivalent, order-unit-level presentation of the same data: a measurement is exactly a channel out of the classical `ι`-outcome system.

Main definitions

  • `Measurement E ι`
  • `Measurement.outcomeDistribution`

4 declarations

definition

Equivalence between measurements and finite families of effects summing to 11

There is an equivalence between a measurement mm on an order-unit space EE with outcome labels in ι\iota and the set of pairs (I,f)(I, f), where I⊆ιI \subseteq \iota is a finite set of outcomes and f:ι→Effect Ef: \iota \to \text{Effect } E is a function assigning an effect to each label, such that the sum of the effects over the finite set II is equal to the identity (order unit) 1∈E1 \in E, i.e., ∑i∈If(i)=1\sum_{i \in I} f(i) = 1.

definition

Outcome distribution of measurement mm in state ss

Given a measurement mm on an order-unit space EE with a finite set of outcomes I⊆ιI \subseteq \iota and a state ss (a unital positive linear functional s:E→Rs: E \to \mathbb{R}), the **outcome distribution** is the probability distribution pp on the set of outcomes II defined by p(i)=s(Ei)p(i) = s(E_i) for each i∈Ii \in I, where EiE_i is the effect corresponding to outcome ii. The resulting values are non-negative and sum to 11, forming an element of the standard simplex ΔI\Delta^I.

theorem

The probability of outcome ii in measurement mm is s(Ei)s(E_i)

Let EE be an order-unit space and mm be a measurement with a finite set of outcomes I⊆ιI \subseteq \iota and associated effects {Ei}i∈I\{E_i\}_{i \in I}. For any state ss (a unital positive linear functional s:E→Rs: E \to \mathbb{R}) and any outcome i∈Ii \in I, the probability of outcome ii in the resulting outcome distribution is equal to the state ss evaluated on the corresponding effect EiE_i: (outcomeDistribution m,s)i=s(Ei) (\text{outcomeDistribution } m, s)_i = s(E_i)

abbrev

Map from states to the outcome distribution of measurement mm

For a finite-outcome measurement mm on an order-unit space EE with a finite set of outcomes I⊆ιI \subseteq \iota, the **outcome map** is the function from the state space S(E)\mathcal{S}(E) to the standard simplex ΔI\Delta^I that assigns to each state ss the probability distribution defined by s↦(s(Ei))i∈Is \mapsto (s(E_i))_{i \in I}, where EiE_i is the effect associated with the outcome i∈Ii \in I.