Is every invariant subspace has an expression of certain form with respect to Primary-Cyclic Decomposition

abstract-algebracommutative-algebralinear algebra

Fix a linear space $V$ over $F$ and a linear operator $T$ on $V$, we know that we can decompose $V$ into Primary-Cyclic subspaces(in the sense of direct sum). Suppose we fix such a Primary-Cyclic decomposition$$V = W_1 \oplus \dots \oplus W_n$$where all the $W_i$ are primary-cyclic(i.e. indecomposable), is it true that every invariant subspace of $T$ is a sum of some of these $W_i$?

We know that since $$primary \& cyclic \iff indecomposable,$$in some sense the Primary-Cyclic decomposition is the "finest" decomposition. The above question I've been thinking about generates naturally from the intuitive expectation of "fine".

Best Answer

It is not necessarily the case that all invariant subspaces can be written as a sum of $W_i$. For instance, consider $$ T = \left[\begin{array}{cc|cc} 0&1&0&0\\0&0&0&0\\ \hline 0&0&0&1\\0&0&0&0 \end{array} \right] $$ with cyclic decomposition $$W_1 = \{(x_1,x_2,0,0):x_1,x_2 \in \Bbb F\}, \quad W_2 = \{(0,0,x_3,x_4):x_3,x_4 \in \Bbb F\}. $$ The invariant subspace $\ker(T) = \{(x_1,0,x_3,0): x_1,x_3 \in \Bbb F\}$ cannot be written as a direct sum of $W_i$.