MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I faced to a bit weird control problem, that is minimize cost functional \begin{equation} J(u) = \int_0^Tg(t,x(t),u(t),\dot u(t))dt \end{equation} subject to \begin{equation} \dot x(t) = f(t,x(t),u(t),\dot u(t)), \quad x(0) = x_0 \end{equation} where $u(t)$ is control, a piecewise continuous function. And it differs from ordinary control problem in presence of $\dot u(t)$ term.

I'd be happy if somebody gave me a hint how this problem could be solved or maybe reduced to an ordinary control problem.

share|cite|improve this question
up vote 1 down vote accepted

Can be reduced to impulse optimal control problem. In the following system \begin{align} &\dot x(t) = f(t,x(t),u(t), v(t)) \newline &\dot u(t) = \nu(t) \end{align} where $\nu(t) = v(t) + \sum\limits_i c_i \delta(t-\tau_i)$ is an impulse control, and $u(t)$ should be considered as another variable.

For further investigation literature on impulse control theory should be looked up. Unfortunately I can't give any references on books in English. But as I know yet there is good theory only for the cases when $f$ is linear in control.

share|cite|improve this answer

Use the standard trick of reducing a higher order ODE to a system of first order ODEs. You should let $v(t)\triangleq\dot u(t)$ be your control, and treat $u(t)$ as another state variable. Then add the equation $\dot u(t) = v(t)$ to your dynamics.

share|cite|improve this answer
That would work if $u(t)$ was abs. continuous, but it is piecewise continuous so it can be represented as $u(t) = u^c(t) + u^d(t)$, where $u^c(t)$ is an absolutely continuous part and $u^d(t)$ is a step function. Then $\dot u(t) = \dot u^c(t)$ and if we put $\dot u(t) = v(t)$ we get \begin{align} &\dot x(t) = f(t,x(t),u^c(t) + u^d(t), v(t)) \newline &\dot u^c(t) = v(t) \end{align} I mean $v(t)$ determines only abs. continuous part of $u(t)$ and we still have to choose $u^d(t)$. It would be ok to consider two dim. control $(v(t), u^d(t))$ if $u^d(t)$ wasn't restricted to be step function. – niyazets Feb 8 '12 at 16:37

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.