add note on StaticArrays in statespace

docs/src/man/numerical.md

@@ -116,4 +116,7 @@ Linear systems with zero-order-hold inputs can be exactly simulated in discrete
 For discrete-time systems, the function [`lsim!`](@ref) accepts a pre-allocated workspace objects that can be used to avoid allocations for repeated simulations.
 
 ### Numerical balancing
-If you are only interested in the simulated outputs, not the state trajectories, you may consider applying balancing to the statespace model using [`balance_statespace`](@ref) before simulating, see the section on [State-space balancing](@ref) above. If the state trajectories are of interest, balancing can still be performed before simulation, and the inverse transformation applied to the state trajectories after simulation.
+If you are only interested in the simulated outputs, not the state trajectories, you may consider applying balancing to the statespace model using [`balance_statespace`](@ref) before simulating, see the section on [State-space balancing](@ref) above. If the state trajectories are of interest, balancing can still be performed before simulation, and the inverse transformation applied to the state trajectories after simulation.
+
+## Static arrays in StateSpace systems
+The special statespace system type [HeteroStateSapce](@ref) can be used to store statespace models with static arrays rather than the default matrix type `Matrix`. See [State-Space Systems](@ref) for more details.