C++17 / Dear ImGui / SDL2 / OpenGL3  —  x86_64 Linux, cross-platform

Stochastic SEIVRD compartmental model with Ornstein-Uhlenbeck noise on β, age-stratified POLYMOD contact mixing, waning immunity, and vaccination sub-compartments. Results are 5-run ensemble averages.

Source (GitHub)  |  Build instructions  |  Model details

Extended SEIVRD compartmental model. Deterministic backbone with stochastic SDE layer (Ornstein-Uhlenbeck on β), integrated via Euler-Maruyama.

Force of infection

Homogeneous:    λ = β · (I + 0.3·E) / N

Age-stratified: λ_i = β · ∑_j C[i,j] · I_j / N_j      (WAIFW)

OU noise:       dβ = θ·(β_0 - β)·dt + σ·β·dW

Integration:    Euler-Maruyama,  dt = 0.5 d (configurable)

Compartments

Symbol	Description	Exits to
S	Susceptible	E (infection), V (vaccination)
E	Exposed / latent	I (at rate σ = 1/incubation)
I	Infectious	R (1−IFR), D (IFR), both at rate γ
H	Hospitalised (flow tracker)	resolved at 1/14 d, extra mort δ
R	Recovered	S (waning ωr)
V	Vaccinated	E (breakthrough, efficacy-reduced), S (waning ωv)
D	Dead (absorbing)	—

Age-stratified extension

8 groups: 0–9, 10–19, 20–29, 30–39, 40–49, 50–59, 60–69, 70+. Contact matrix C[i,j] defaults to POLYMOD European average (Mossong 2008, Prem 2017). IFR follows a COVID-like age gradient as default; all values user-editable. Vaccination budget distributed proportionally to group-level uptake fractions, capped at configurable coverage ceiling per group.

Ensemble strategy

N_RUNS=5 independent Euler-Maruyama integrations from distinct MT19937-64 seeds. Time series averaged pointwise. 10th and 90th percentile bands computed per compartment. Removes single-seed luck without masking structural uncertainty.

Parameters

Parameter	Default	Description
β	0.3	Baseline contact × transmission probability (per day)
σ	0.2	1 / incubation period (E→I rate)
γ	0.1	1 / infectious period
IFR	0.01	Infection fatality ratio
hosp_rate	0.05	Fraction of I requiring hospitalisation
ωr	0.003	Waning rate, recovered (~1 yr)
ωv	0.002	Waning rate, vaccinated
vax_rate	0.0	Daily fraction of S vaccinated
vax_eff	0.85	Vaccine efficacy against infection
season_amp	0.0	Cosine forcing amplitude on β
σOU	0.05	OU noise amplitude (fraction of β)
θOU	0.1	OU mean-reversion speed

CMake 3.18+, C++17 compiler, OpenGL. SDL2 and Dear ImGui fetched automatically via FetchContent if not present on the system.

# Linux (primary target)
sudo apt install libsdl2-dev libgl-dev   # optional; FetchContent fallback
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build -j$(nproc)
./build/clades

# Windows / macOS - identical
cmake -B build -DCMAKE_BUILD_TYPE=Release
cmake --build build

Source files

File	Role
model.hpp / model.cpp	Homogeneous SEIVRD integrator, ensemble runner, all core types
age_strat.hpp / age_strat.cpp	Age-stratified integrator, POLYMOD defaults, per-group ensemble
io.hpp / io.cpp	INI param save/load for base and age params
main.cpp	Dear ImGui frontend, all UI panels, SDL2/OpenGL loop

Auto-fetched dependencies

Library	Version	Purpose
Dear ImGui	1.90.8	Immediate-mode GUI
ImPlot	0.16	Plot widgets inside ImGui
SDL2	2.30.2	Window, OpenGL context, input