Model classes satisfying the square-integrability conditions #
The square-integrable relaxed regime asks for three L² conditions: on the
random test bound, on the random intensity bound along the visited flow
segments, and on the scaled reciprocal of the random lower bound for the
holding times. This module discharges the last two for the model classes in
which they are automatic, and assembles the resulting statements.
- A uniform intensity bound over the state space is a constant intensity bound along the trajectory, and constants are square integrable over a finite measure.
- A deterministic minimal holding time is a constant lower bound, and its scaled reciprocal is again constant, hence square integrable.
Combining the two leaves a single genuine hypothesis, square integrability of the path-level test bound. The resulting statement strictly extends the bounded regime, which needed a test function bounded uniformly over the whole state space: here the test function only has to be bounded along the visited trajectory by a square-integrable random variable.
The two automatic conditions #
A uniform bound on the total intensity is a constant intensity bound along every visited flow segment.
A constant intensity bound is square integrable over a finite measure.
A deterministic strictly positive minimal holding time is a constant random lower bound whose scaled reciprocal is square integrable.
Uniform intensity and deterministic minimal holding time #
The two product hypotheses for a uniform model. With a uniform bound on the total intensity and a deterministic strictly positive minimal holding time, the two product hypotheses of the path-level regime follow from square integrability of the path-level test bound alone.
Martingale property for a uniform model with a square-integrable test bound. Under a uniform intensity bound and a deterministic strictly positive minimal holding time, the occupation-centered reset process is a martingale for the full-information filtration as soon as the test function is bounded along the trajectory by a square-integrable random variable, with the deterministic-cap increment law as the only remaining structural input.
Dynkin-type identity for a uniform model with a square-integrable test bound.
A uniformly bounded test function is the degenerate case #
With a uniformly bounded test function the square-integrability hypothesis is automatic, so the uniform model recovers the bounded regime.