Skip to content

Chain ​

  • 3 state
  • 1 control
  • Mayer cost
  • fixed final time
  • 5 boundary
  • OptimalControl valid
  • OptimalControl, scalarized valid
  • JuMP valid
  • ModelingToolkit valid

The Hanging Chain problem is a classical benchmark in optimal control. It consists of moving a chain from a given initial horizontal position to a target horizontal position while controlling the horizontal velocity of the chain. The objective is to reach the final configuration in a way that minimises the vertical displacement of the chain. This problem is widely used to test trajectory optimisation and direct transcription methods for nonlinear optimal control.

Mathematical formulation ​

The problem can be stated as

Qualitative behaviour ​

The optimal control trajectory exploits the nonlinear coupling between horizontal and vertical motion:

  • The state directly follows the control input (horizontal velocity).

  • The state evolves depending on , which introduces nonlinear dynamics in the vertical motion.

  • The state measures the chain extension and grows with the magnitude of .

The control typically balances horizontal movement to minimise the vertical displacement at the final time.

Characteristics ​

  • Nonlinear dynamics.

  • Minimum vertical displacement objective.

  • Serves as a benchmark for trajectory optimisation and direct transcription methods in nonlinear optimal control.

References ​

  • More, J. J., & Munson, T. S. (2000). The Hanging Chain Problem as an Optimal Control Problem. mcs.anl.gov/~more/cops/bcops/chain.html This benchmark explicitly formulates the Hanging Chain (catenary) as an optimal control problem, including direct transcription to an NLP. It is widely used as a test case for solver performance in AMPL, MINOS, and other tools.

  • Dolan, E. D., & More, J. J. (2001). Benchmarking Optimisation Software with COPS 3.0. Technical Report ANL/MCS-TM-246, Argonne National Laboratory. mcs.anl.gov/~more/cops The COPS benchmark collection officially includes the Hanging Chain problem as one of its core examples. It provides comprehensive problem formulation, discretisation strategies, and solver comparison results.

  • Rutquist, P. E., & Edvall, M. M. (2009). Hanging Chain problem example in PROPT MATLAB Optimal Control Software. Included in the PSOPT distribution’s example suite, as credited by PSOPT’s list of examples (psopt.net). This demonstrates a practical implementation of the Hanging Chain problem via direct transcription using MATLAB-based optimal control software.

  • Huygens, C. (1690). Horologium Oscillatorium. Paris: F. Muguet. This foundational work contains one of the earliest studies of the hanging chain (catenary) curve. Huygens, along with Leibniz and Johann Bernoulli, contributed to the historical derivation of the catenary equation, which underlies modern optimal control formulations of the problem.

Parameters ​

ParameterDefault valueUnitMeaning
t00–initial time
tf1–final time
L4–length of the chain, the final value of x₃
a1–initial value of x₁
b3–final value of x₁
x₂_t00–initial value of x₂
x₃_t00–initial value of x₃

Every modeling takes other values with the keyword parameters, merged with these: chain(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.

ModelingSchemeGridReference objectiveSolvesDistinct solutionsConfigurations agree
JuMP:trapeze5005.0685777981yes, within 5.3e-16
OptimalControl:trapeze5005.06857779181yes, within 5.2e-13
OptimalControl, scalarized:trapeze5005.0685777931yes, within 0