Quick start

A first model

The smallest non-trivial continuous-time saddle: a stable state x (a decaying transition from an initial condition) and an unstable jump y (anchored at its terminal steady state of zero):

var(state) x;
var(jump)  y;

model;
  diff(x) = -x;
  diff(y) = y;
end;

initval;
  x = 1;
end;

simulate(T=5, N=20);

Save it as saddle.mod. The model has one state (predetermined, pinned by initval) and one jump (forward-looking, pinned at the terminal steady state by the solver). The model; block lists the equations; simulate(...) requests a perfect-foresight solve over [0, 5] discretised on a 20-interval grid.

From the command line

$ continuo saddle.mod
continuo: wrote 21 rows to saddle.csv

This writes one row per grid point (header t, x, y) into saddle.csv. Override the horizon, grid resolution, or output path:

$ continuo saddle.mod -T 10 -N 100 -o /tmp/saddle.csv

See Command-line interface for the full CLI reference.

From Python

import continuo

model = continuo.parse("saddle.mod")
sol = model.simul()                 # uses T, N from the simulate command
print(sol["x"][0], sol["x"][-1])    # 1.0, ~0.0067

ss = model.steady_state()           # the trivial steady state x=y=0

The Model returned by parse() exposes the parsed model and its solvers; Solution is the result of simul(), with NumPy arrays for the time grid and the solved path. See Python API for the full reference.

A richer example

The shipped Ramsey/RBC model in examples/rbc/ is the natural next step. It illustrates:

• two state variables (capital K and productivity A),

• a jump (consumption C) and an algebraic variable (output Y),

• an analytical steady_state_model block,

• shock paths and surprises in the shocks block,

• factoring the shared model out of multiple scenarios with the macroprocessor (@#include "common.mod").

See Worked examples for the full list of worked-out models, and The .mod language for the reference on each construct used.