Let $F, G: \mathbb{R}^3\rightarrow{}\mathbb{R}$ be two continuously differentiable functions and let $a\leq b \leq c$. I want to know if there exists some known method to find a function that maximizes the functional $J(P):=\int_a^b F(q,P(q),P'(q))dq+ \int_b^c G(q,P(q),P'(q))dq$ between all the continuously differentiable functions $P: \mathbb{R}\rightarrow{}\mathbb{R}$ satisfying $P(a)=0$.

  • 1
    $\begingroup$ Is it the case that $F(b,P(b),P'(b))=G(b,P(b),P'(b))$? If so and their derivatives also match at $b$ then you can define a piecwise $C^1$ function and use the Euler Lagrange equations. $\endgroup$ – Basel J. Jul 10 at 19:45
  • $\begingroup$ Comment to the post (v1): Does $P$ satisfies any conditions at $b$ and/or $c$? $\endgroup$ – Qmechanic Jul 11 at 11:02
  • $\begingroup$ Yes: I need $F(b,P(b),P'(b))=0=G(c,P(c),P'(c))$ $\endgroup$ – asd Jul 12 at 19:54
  • $\begingroup$ Those seem not necessary. Other conditions? $\endgroup$ – Qmechanic Jul 13 at 8:20

This is known as a multiphase optimal control problem, and there is software available online for this kind of problems. Examples are DIDO (matlab, proprietary), GPOPS-II (matlab, proprietary), PROPT(matlab, proprietary - i think), ICLOCS2 (matlab, free) and PSOPT (C++, free).

They all use pseudospectral methods, which are great to feed a computer, but not very practical to solve by hand. I don't know any method to solve analytically. However, I am certain that if $F$ and $G$ are not identical, then the Hamiltonian will (most likely) have a jump discontinuity at q = b.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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