I have read about matching networks, authors usually will describe a target impedance to be presented, to say, the output of an amplifier, and then they design the matching network. To widen the bandwith they usually create a matching network with more sections, thus, lowering the general Q and presenting the same impedance to a wider range of frequencies.
But my question is this: What's the point on presenting the same impedance over a wide range of frequencies if the S-parameters (and thus the required load/source to be presented to the amp) change with frequency? For example the optimum source impedance for LNA design is not the same at different frequencies, in a power amplifier the optimum load varies with frequency, so whats the point on designing a matching network that presents the same load over a wide range of frequencies?, instead, shouldnt a network be designed to "track" those impedance changes, thus presenting the ideal source/load required for each frequency?, how is a network like that designed?