Vanderpol
- 2 state
- 1 control
- Lagrange cost
- fixed final time
- 2 boundary
- OptimalControl valid
- OptimalControl, scalarized valid
- JuMP valid
- ModelingToolkit valid
The Van der Pol oscillator control problem is a benchmark in nonlinear optimal control. It models a two-dimensional oscillator with nonlinear damping and a control input. The state vector is
Mathematical formulation
The problem can be stated as
Qualitative behaviour
The optimal control
regulates the nonlinear oscillations to minimise both state excursions and control effort. Due to the quadratic cost, the control is typically smooth, without bang–bang or singular arcs.
Serves as a standard test for direct transcription methods and numerical solvers for nonlinear optimal control problems.
Characteristics
Nonlinear dynamics with cubic damping.
Quadratic cost on both states and control.
Useful benchmark for testing optimal control algorithms and integration schemes.
References
Dymos Documentation. The Van der Pol Oscillator.. Provides an implementation of the Van der Pol oscillator in an optimal control framework, illustrating problem setup, discretisation, and solution.
mintOC. Van der Pol Oscillator. Presents the Van der Pol optimal control problem with numerical results, highlighting cost function and control profiles.
James, E.M. (1974). Time Optimal Control and the Van der Pol Oscillator. IMA Journal of Applied Mathematics, 13(1), 67–78. Discusses theoretical aspects of time-optimal control for the Van der Pol oscillator, including analysis of control strategies.
Van Dooren, R. (1987). Numerical Study of the Controlled Van der Pol Oscillator in Optimal Control. Numerische Mathematik, 51(4), 471–485. Explores numerical methods for solving the Van der Pol optimal control problem, providing insights into solution behavior.
Parameters
| Parameter | Default value | Unit | Meaning |
|---|---|---|---|
t0 | 0 | – | initial time |
tf | 2 | – | final time |
ω | 1 | – | frequency of the oscillator |
ε | 1 | – | coefficient of the nonlinear damping |
x₁_t0 | 1 | – | initial value of x₁ |
x₂_t0 | 0 | – | initial value of x₂ |
Every modeling takes other values with the keyword parameters, merged with these: vanderpol(backend; parameters=(t0=…,)).
Reference objective
The best objective among the solves that converged, in each modeling, on the default grid of the modeling: the modelings are independent, each has its own reference. See how the references are computed.
| Modeling | Scheme | Grid | Reference objective | Solves | Distinct solutions | Configurations agree |
|---|---|---|---|---|---|---|
| JuMP | :trapeze | 500 | 1.047811552 | 8 | 1 | yes, within 0 |
| OptimalControl | :trapeze | 500 | 1.047811552 | 18 | 1 | yes, within 4.0e-15 |
| OptimalControl, scalarized | :trapeze | 500 | 1.047811552 | 3 | 1 | yes, within 2.5e-15 |