Frequently asked questions

How do I create a modified distribution?

Each modifier is a plain function taking a distribution and returning a distribution. The printed result below shows each wrapper displaying the distribution it wraps:

using ModifiedDistributions, Distributions

[affine(LogNormal(1.5, 0.5); scale = 2.0, shift = 1.0),
    weight(Normal(2.0, 1.0), 10.0),
    thin(Gamma(2.0, 1.0), 0.3),
    cumulative(Gamma(2.0, 1.0)),
    modify(Weibull(1.5, 2.0), 0.5)]

5-element Vector{ModifiedDistributions.AbstractModifiedDistribution{Distributions.Univariate, Distributions.Continuous}}:
 Affine(Distributions.LogNormal{Float64}(μ=1.5, σ=0.5))
 Weighted(Distributions.Normal{Float64}(μ=2.0, σ=1.0))
 Transformed(Distributions.Gamma{Float64}(α=2.0, θ=1.0))
 Transformed(Distributions.Gamma{Float64}(α=2.0, θ=1.0))
 Modified(Distributions.Weibull{Float64}(α=1.5, θ=2.0), 0.5; link=LogLink)

What are the three weight scenarios for weight?

2. A fixed constructor weight: weight(d, 10.0) multiplies logpdf by 10.

3. An observation-time weight: weight(d) stores missing and expects joint observations (value = x, weight = w).

4. Vectorised weights: weight(d, [3, 1, 4]) builds a Product of weighted components for vector observations.

Constructor and observation weights combine by multiplication when both are present.

In every form the result stays a real, samplable Distributions.jl distribution. See the Weighted likelihoods tutorial's Weighted distributions stay samplable section for why that matters in a PPL.

Why does a zero or missing weight give -Inf rather than NaN?

0 * logpdf would be NaN at an out-of-support value (logpdf = -Inf), which poisons a sampler; weight short-circuits a zero or missing weight straight to -Inf instead. See the Weighted likelihoods tutorial's Why zero and missing weights give -Inf section for the full reasoning, including why it keeps automatic differentiation well defined.

Why doesn't thin change logpdf?

thin and cumulative are forward-series transforms: they carry an operation for a count series a downstream layer (for example a convolution engine) produces. The distribution itself is unchanged — logpdf, rand, cdf and everything else delegate to the inner distribution. If you want thinning that changes the density, that is a different operation and lives with the convolution layer that owns the series.

What is the difference between get_dist and get_dist_recursive?

get_dist removes one layer of wrapping; get_dist_recursive keeps unwrapping until it reaches a distribution with no more layers. See the Unwrapping section of the Getting started overview for a worked example.

Can I combine modifiers in any order?

Yes. Each modifier is a thin wrapper around whatever you pass it, so weight(affine(d; scale = 2), 10) and affine(weight(d, 10); scale = 2) both work. Order matters for meaning, not validity: modify the distribution first, then weight the resulting likelihood term, unless you have a reason to do otherwise.

Which hazard modifications does modify support?

The effect is a scalar, a callable effect(t), or a per-bin vector, and the path is chosen by the base and the link (see the Getting started overview and the modifier pipeline tutorial for the closed-form maths).

Base	Effect	Link	Behaviour
Continuous	Scalar	log	Closed form: proportional hazards, survival raised to exp(effect).
Continuous	Scalar, negative	identity	Closed form: hazard clamped to max(h(t) + β, 0), survival reconstructed exactly from the base cumulative hazard between clamp knots (no quadrature). A non-monotone base hazard can go sub-stochastic (defective) under a strong negative effect, and quantile/rand above the total mass then throw.
Discrete	Per-bin vector	any (including :logit or a user callable)	Closed form: reshapes each delay bin's reporting hazard on the link scale, reconstructing the PMF exactly — the epinowcast discrete-time reporting hazard.
Continuous	Callable effect(t)	any	Deferred: a time-varying hazard has no closed-form cumulative hazard, so it needs numeric integration. Tracked in issue #77 part (b); such a Modified constructs, but evaluating it throws until the path lands.

The epinowcast reference-by-report expected-count matrix layer stays upstream in CensoredDistributions.jl.

Does this work with composed distribution chains?

Yes. Loading ComposedDistributions.jl activates a package extension that lets the modifier verbs apply to a Sequential chain's one observed scalar (a Parallel has no single observed scalar, so the verbs are not defined for it). See the Modifiers across composed chains tutorial for why a chain has exactly one observed scalar and a worked example.

How do I cite ModifiedDistributions?

See the citation section of the README.