Table 1. An overview of the parameters and variables.

symbol	unit	description
Nt	-	number of recrudescence events at time t post treatment interruption
Vt	copies mL−1	VL at time t post treatment interruption
Ti	d	time of the i-th successful reactivation event.
g	d−1	exponential growth rate of the VL
v0	copies mL−1	average initial viral concentration caused by a single successful reactivation
λ	d−1	recrudescence rate, the number of latently infected cells that successfully reactivate per day.
K(θ)	-	cumulant-generating function of the viral load Vt.
κ1	copies mL−1	first cumulant, equal to the expectation of the VL ($\mathbb{E}\left[V_t\right]$).
κ2	copies2mL−2	second cumulant, equal to the variance of the VL (Var[Vt]).
${\tilde{V}}_t$	copies mL−1	conditionally deterministic approximation of the viral load process Vt.
ℓ	copies mL−1	limit of detection of the VL assay.
τ	d	viral rebound time, satisfies the equation Vτ = ℓ
S(t)	-	fraction of subjects in remission at time t post treatment interruption
f(t; λ, g, v0, ℓ)	d−1	approximation of the probability density function of the rebound time τ
${\widehat{T}}_1$	d	first recrudescence time extrapolated from the rebound time using the growth rate and the initial VL, assuming simple exponential growth.
tART	d	days post infection that ART was initiated
Gi	d−1	random, clone-specific exponential growth rate.
σG	d−1	standard deviation of the growth rates Gi of the clones comprising the SIV reservoir

See also Table 2 for additional parameters of the Bayesian mixed-effects model.