Two results on exact boundary control of parabolic equations (Q1056900): Difference between revisions

The first result is on the time-dependent boundary control system \[ (1)\quad u_ t=u_{xx}+q(t,x)u\quad (0\leq x\leq L,\quad t\geq 0), \] \[ (2)\quad u(0,x)=u_ 0(x)\quad (0\leq x\leq L), \] \[ (3)\quad u(t,0)=0,\quad u(t,L)=f(t)\quad (t\geq 0). \] Using in a very clever way seemingly unrelated results of D. Colton, the author shows that (1)-(2)- (3) is null controllable in any time \(T>0\), that is, for any \(u_ 0(x)\) in \(L^ 2(0,L)\) there exists a control f(t) in \(L^ 2(0,T)\) such that the solution of (1)-(2)-(3) satisfies \(u(T,x)=0\) (0\(\leq x\leq L).\) To the reviewer's knowledge, this is the first (and only) result on null controllability of time-dependent boundary control systems; apart from its intrinsic importance, the result will perhaps be useful in the theory of optimal control problems for nonlinear boundary control systems. The second result is on ''cost of control'' for the heat equation. Let \(C_ T\) be the operator that associates with any \(u_ 0(x)\) in \(L^ 2(0,L)\) the control of minimum norm f(t) in \(L^ 2(0,T)\) driving it to zero in time T. A conjecture dating back to the late sixties is that \((4)\quad \| C_ T\| =O(\exp (1/T))\) as \(T\to 0\). The author proves this conjecture in a remarkably simple way. His estimate is made much more interesting by a recent result which shows that (4) is sharp. Thus one can conclude that for the heat equation (and most likely for any parabolic equation) the cost of control becomes prohibitive as \(T\to 0\), a situation that contrasts sharply with that for hyperbolic equations (where null controllability in small time is anyway impossible, due to finite velocity of propagation).