Let M be a forward-shift-invariant subspace and N a backward-shift-invariant subspace in the Hardy space H 2 on the bidisc. We assume that H 2 = N ⊕ M. Using the wandering subspace of M and N , we study the relations between M and N . Moreover we study M and N using several natural operators defined by shift operators on H 2 .2000 Mathematics subject classification: primary 47A15, 46J15; secondary 47A20. Keywords and phrases: invariant subspace, two variable Hardy space, wandering subspace.
IntroductionLet 2 be the torus, that is, the Cartesian product of two unit circles in C . Let p = 2 or p = ∞. The usual Lebesgue spaces, with respect to the Haar measure m on 2 , are denoted by L p = L p ( 2 ), and H p = H p ( 2 ) is the space of all f in L p whose Fourier coefficientsf ( j, ) = 2 f (z, w)z jw dm (z, w) vanish if at least one of j and is negative. Then H p is called the Hardy space. As 2 = z × w , H p ( z ) and H p ( w ) denote the one-variable Hardy spaces. Let P H 2 be the orthogonal projection from L 2 onto H 2 . For φ in L ∞ , the Toeplitz operator T φ is defined by T * z N ⊂ N and T * w N ⊂ N . Let P M and P N be the orthogonal projections from H 2 onto M and N , respectively. In this paper, we assume that M ⊕ N = H 2 , that is, P M + P N = I where I is the identity operator on H 2 . LetSuppose thatIt is known [4] that AB| M = V and B A| N = S. Guo and Yang [3] showed that AB is Hilbert-Schmidt under some mild conditions. In this paper, we study M or N when A, B, AB or B A is of finite rank. Izuchi and Nakazi For a forward-shift-invariant subspace M, putThese are called wandering subspaces for M. In this paper, [·] denotes the closed span. For a backward-shift-invariant subspace N , with M = H 2 N , putThese are called wandering subspaces for N . In Section 2 we decompose and study M and N using the wandering subspaces M 1 , M 2 , N 1 and N 2 . In Section 3 we study M and N when A or B is of finite rank. For an operator K , r (K ) denotes the rank of K . In Section 4 we show that r (AB) = dim N 1 ∩ N 2 in general, and r (B A) = dim M 1 ∩ M 2 under some mild conditions.In this paper, for a bounded linear operator X on H 2 , ranX = X H 2 and ker
Wandering subspacesLet M be a forward-shift-invariant subspace, and N be a backward-shift-invariant subspace with Invariant subspaces in the bidisc and wandering subspaces 369andIn the case of one variable,is an inner function of one variable. THEOREM 1. Let N be a backward-shift-invariant subspace and M = H 2 N .z , then by definition z n f ∈ N for any n ≥ 1, and hence f is orthogonal to [Hence z 2 f ∈ N . By repeating the same argument, we can show that z n f belongs to N for any n ≥ 1. This implies (2). 2 COROLLARY 2. Let N be a backward-shift-invariant subspace. Conversely, if M = q 2 H 2 then M 1 = q 2 H 2 ( w ), and so N 1 = T * z M 1 = 0. Part (2) is clear by (2) of Theorem 1.2By (1) of Theorem 1, both M 1 and M 2 are cyclic subspaces for T z and T w : that is,It may happen that In general, N 0 may not be a cyclic subspace because N 0 = [0] may ...