In this note$^1$ page 4 the author states that suppose $a_i$ acts on the whole vector space $V$ where $a_i$ is the annihilation operator of fermions. And since $[a_i^\dagger a_i,a^\dagger_j a_j]=0,\forall i,j$, and $a_i^\dagger a_i$ are hermitian operators, so we can construct a basis for them as $\left( {a_1}^{\dagger} \right) ^{\alpha _1}...\left( a_{n}^{\dagger} \right) ^{\alpha _n}|vac\rangle $ where $\alpha _i=0$ or $1$. Then the author states that the constructed vector space $W$ is a subspace of $V$, but I wonder that hermitian operators commuting implies that we have a common complete orthogonal basis, so how can $W$ be only a subspace of $V$?
1. "The Fermionic canonical commutation relations and the Jordan-Wigner transform" By Michael A. Nielsen, 2005