Continuo 0.0.4-dev

continuo solves continuous-time, perfect-foresight deterministic dynamic models in the spirit of Dynare. A .mod file declares the endogenous and exogenous variables, the model equations (as differential equations in continuous time, written with diff(x)), the steady state and the exogenous shock paths; the toolbox parses it, discretises the two-point boundary-value problem by collocation (Crank–Nicolson by default, or a higher-order Gauss / Radau IIA / Lobatto IIIA family), and solves the stacked nonlinear system with Newton’s method. Anticipated changes, surprises (multi-segment beliefs) and occasionally-binding constraints are all supported through the same machinery.

The toolbox has three faces:

• the .mod surface language (the canonical input format),

• a Python API (continuo) for use in scripts and notebooks,

• a command-line wrapper continuo that reads a .mod file and writes the solved path to CSV.

This manual is the reference for all three. The Quick start shows the smallest end-to-end use; the The .mod language documents every block and built-in of the surface language; Python API is the Python reference (generated from docstrings); Discretisation schemes documents the discretisation schemes (Crank–Nicolson and the higher-order collocation families); Time grids and adaptive refinement documents the time grid and adaptive refinement; Linear solvers documents the pluggable linear backends and how to choose one; Steady-state solvers does the same for the nonlinear steady-state algorithms; Worked examples is the index of the worked-out example models that ship with the source tree.

Getting started

• Installation
  • Requirements
  • From a release
  • From a clone
  • Optional extras
  • Linear-solver backends
  • Verifying the install
• Quick start
  • A first model
  • From the command line
  • From Python
  • A richer example

Reference

• The .mod language
  • Declarations
  • Domain constraints
  • The model block
  • The steady_state_model block
  • Initial conditions: initval and initial_guess
  • The shocks block
  • Commands
  • Expressions
  • The macroprocessor
• Python API
  • Entry points
  • The Model object
  • The Solution object
  • Diagnostics
  • Exceptions
• Discretisation schemes
  • Choosing a scheme
  • The families
  • How it works
  • Accuracy
• Time grids and adaptive refinement
  • Shock-aligned nodes
  • Adaptive refinement
• Linear solvers
  • Choosing a backend
  • Presets
  • auto routing
  • The backends
  • Fine control
  • Guard-rails
  • Diagnostics
  • Benchmarks
• Steady-state solvers
  • Choosing a solver
  • Presets
  • auto routing
  • The backends
  • Solver options
  • Fine control
  • Domain constraints
  • Diagnostics
• Command-line interface
  • Synopsis
  • Positional argument
  • Options
  • Output format
  • Exit status
  • Examples
  • Programmatic equivalent
• Worked examples
  • Saddle-path models (one state + one jump)
  • Pure forward-looking models (jumps only, no state)
  • Pure initial-value problems (states only, no jump)

Indices

• Index

• Module Index

• Search Page