Stopping times, stopped processes and stopped values #
Definition and properties of stopping times.
Main definitions #
MeasureTheory.IsStoppingTime: a stopping time with respect to some filtrationfon a measurable spaceΩis a functionτ : Ω → WithTop ιsuch that for alli : ι, the preimage of{j | j ≤ i}alongτisf i-measurableMeasureTheory.IsStoppingTime.measurableSpace: the σ-algebra associated with a stopping time
Main results #
ProgMeasurable.stoppedProcess: the stopped process of a progressively measurable process is progressively measurable.memLp_stoppedProcess: if a process belongs toℒpat every time inℕ, then its stopped process belongs toℒpas well.
Implementation notes #
For a filtration on a type ι, we define stopping times as functions from the measurable space Ω
to WithTop ι, which allows stopping times that can take an infinite value, represented by
⊤ : WithTop ι.
This means that if we have a process X : ι → Ω → β and a stopping time τ : Ω → WithTop ι, then
to consider the value of X at the stopping time τ ω, we need to write X (τ ω).untopA ω,
in which (τ ω).untopA is the value of τ ω in ι if τ ω ≠ ⊤ and some arbitrary value if
τ ω = ⊤.
While indexing would be more convenient if we defined stopping times as functions from Ω to ι,
this would prevent us from using stopping times as in standard mathematical literature, where a
typical example of stopping time is the first time an event occurs, which may never happen.
Consider for example the first time a coin lands heads when flipping it infinitely many times:
this is almost surely finite, but possibly infinite. We could also not use a function Ω → ι with
arbitrary value for the infinite case, because this would be incompatible with the stopping time
property.
Tags #
stopping time, stochastic process
Stopping times #
A stopping time with respect to some filtration f is a function
τ such that for all i, the preimage of {j | j ≤ i} along τ is measurable
with respect to f i.
Intuitively, the stopping time τ describes some stopping rule such that at time
i, we may determine it with the information we have at time i.
Equations
- MeasureTheory.IsStoppingTime f τ = ∀ (i : ι), MeasurableSet {ω : Ω | τ ω ≤ ↑i}
Instances For
Auxiliary lemma for MeasureTheory.IsStoppingTime.measurableSet_lt.
The associated σ-algebra with a stopping time.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Alias of MeasureTheory.IsStoppingTime.measurableSet_eq_stopping_time.
Stopped value and stopped process #
Given a map u : ι → Ω → E, its stopped value with respect to the stopping
time τ is the map x ↦ u (τ ω) ω.
Equations
- MeasureTheory.stoppedValue u τ ω = u (τ ω).untopA ω
Instances For
Given a map u : ι → Ω → E, the stopped process with respect to τ is u i ω if
i ≤ τ ω, and u (τ ω) ω otherwise.
Intuitively, the stopped process stops evolving once the stopping time has occurred.
Equations
- MeasureTheory.stoppedProcess u τ i ω = u (min (↑i) (τ ω)).untopA ω
Instances For
The stopped process of a strongly adapted process with continuous paths is strongly adapted.
If the indexing order has the discrete topology, then the stopped process of a strongly adapted process is strongly adapted.
Filtrations indexed by ℕ #
Given stopping times τ and η which are bounded below, Set.piecewise s τ η is also
a stopping time with respect to the same filtration.