A bounded operator on a Hilbert A bounded operator on a Hilbert space which has no square root.

******************************

Suppose that T is the operator T(x1, x2, ¼) = (x2, x3, ¼) on l2 ( in fact T is the adjoint of the unilateral shift operator). If T has a square root S, then S2 = T and KerS Í KerT = C x1 in which x1 = (1, 0, 0, ¼). Since T is not one to one we conclude that S is not one to one. So that KerS = C x1.  T is surjective, hence S is onto. So there exists an element h such that Sh = x1. Since Th = S2h = Sx1 = 0, we have h = lx1 for some l Î C and hence x1 = Sh = lS x1 = 0, a contradiction.

Comment. There is an open subset of L(H) consisting of invertible operators with no square roots.
Ref.
J.B. Conway and B.B. Morrel, Roots and logarithms of bounded operators on Hilbert spaces, J. Funct Anal, 70(1987), 171-193.


File translated from TEX by TTH, version 2.70.
On 22 Feb 2001, 00:21.