Brachistochrone
- 3 state
- 1 control
- 1 variable
- Mayer cost
- free final time
- 1 control box
- 1 variable box
- 5 boundary
- OptimalControl valid
- OptimalControl, scalarized valid
- JuMP valid
The Brachistochrone problem is a classical benchmark in the history of calculus of variations and optimal control. It seeks the shape of a curve (or wire) connecting two points
Mathematical formulation
The problem can be stated as a free final time optimal control problem:
where
Qualitative behaviour
This problem is a classic example of minimum time control with nonlinear dynamics. Unlike the shortest path problem (a straight line), the optimal trajectory balances the need to minimize path length with the need to maximize velocity.
The analytical solution to this problem is a cycloid—the curve traced by a point on the rim of a circular wheel as the wheel rolls along a straight line without slipping. Specifically:
Energy Conservation: Since the system is conservative and frictionless, the speed at any height
is given by (assuming start from rest). This implies that maximizing vertical drop early in the trajectory increases velocity for the remainder of the path.Concavity: The optimal curve is concave up. It starts steeply (vertical tangent if
) to gain speed quickly, then flattens out near the bottom before potentially rising again to reach the target.Singularity: At the initial point, if
, the control is theoretically singular as the bead has no velocity to direct. Numerical solvers often require a small non-zero initial velocity or a robust initial guess to handle this start.
The control
Characteristics
Nonlinear dynamics involving trigonometric functions of the control.
Analytical solution is a Cycloid.
Historically significant as the first problem solved using techniques that evolved into the Calculus of Variations.
References
- Bernoulli, J. (1696). Problema novum ad cujus solutionem Mathematici invitantur.
Acta Eruditorum. The original publication posing the challenge to the mathematicians of Europe, solved by Newton, Leibniz, L'Hôpital, and the Bernoulli brothers.
- Sussmann, H. J., & Willems, J. C. (1997). 300 years of optimal control: from the brachystochrone to the maximum principle.
IEEE Control Systems Magazine. doi.org/10.1109/37.588108 A comprehensive historical review linking the classical Brachistochrone problem to modern optimal control theory and Pontryagin's Maximum Principle.
- Dymos Examples: Brachistochrone. OpenMDAO/Dymos
The numerical implementation serving as the basis for the current benchmark formulation.
Parameters
| Parameter | Default value | Unit | Meaning |
|---|---|---|---|
g | 9.80665 | m/s² | gravitational acceleration |
t0 | 0.0 | s | initial time |
x0 | 0.0 | m | initial horizontal position px |
y0 | 10.0 | m | initial vertical position py |
v0 | 0.0 | m/s | initial speed |
xf | 10.0 | m | final horizontal position px |
yf | 5.0 | m | final vertical position py |
u_min | -1.57 | rad | lower bound of the angle u |
u_max | 1.57 | rad | upper bound of the angle u |
tf_l | 0.1 | s | lower bound of the final time |
tf_u | 20.0 | s | upper bound of the final time |
Every modeling takes other values with the keyword parameters, merged with these: brachistochrone(backend; parameters=(g=…,)).
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 | 1.802932061 | 8 | 1 | yes, within 8.4e-08 |
| OptimalControl | :trapeze | 500 | 1.802932061 | 16 | 2 | no |
| OptimalControl, scalarized | :trapeze | 500 | 1.802932159 | 3 | 1 | yes, within 0 |
In OptimalControl with :trapeze, the solves reached 2 distinct solutions: besides the best one, 1.843496126 (gap 2.2e-02, 2 solves). See the results of the modeling for which configurations reached which.