The [[Frequency Response Function]] of some [[Signal Filter]] is a [[Function]] [[Invariant]] of the [[Time Domain]] $H(\omega)$. This function comes about from the process of assuming our input $x_{t}$ is of the form of $e^{i\omega t}$ and getting filter output $y_{t}=H(\omega)e^{i\omega t}$, where $H(\omega)\in \mathbb{C}$. $\huge \begin{align} y_{t} &= H(\omega) e^{i\omega t} \\ H(\omega) &= r_{\omega}e^{i\theta_{\omega}} \\ \end{align} $ >[!note] > This page uses the second example from [[Finite Impulse Response Filters]]. ## Magnitude and Phase Response The [[Frequency Response Function|magnitude response]] is defined as the [[Absolute Value]] of $H(\omega)$ $\huge \begin{align} H(\omega) &= \underbrace{\left| H(\omega) \right|}_{\text{Magnitude Response}}e^{i \overbrace{\theta \omega}^{\text{Phase Response}}} \\ \end{align} $ Where $\theta$ is the [[Phase]]. $H(\omega)$ is a function of a [[Real Numbers|Real Number]] *or* can be "embedded" into the [[Complex Plane]] on the [[Unit Circle]], ie with inputs $e^{i\omega}$ with $0\leq\omega\leq{2}\pi$. >[!example] Simple Observations on the unit circle and [[Frequency Response Function|Magnitude Response]] > $\huge \begin{align} > y_{t} = x_{t}-a_{1}x_{t-1}, \, 0<a_{1}<1, a_{1} \approx 1 > \end{align} $ > $\huge \begin{align} > \mathcal{H}(z) &= 1-a_{1} z^{-1} \\ > H(\omega) &= \frac{{e^{i \omega }-a_{1}}}{e^{i\omega}} \\ > > \left| H(\omega) \right| &= \left| \frac{{e^{i \omega }-a_{1}}}{e^{i\omega}} \right| \\ > |e^{i\omega} | &= 1 \\ > |H(\omega)| &= \left| e^{i\omega} -a_{1} \right| > \end{align} $ > > Note that $t$ is in [[Sampling|samples]]. > > $\huge \begin{align} > \min|H(\omega)| &= 1-a_{1} \\ > \max|H(\omega)| &= 1+a_{1} \\ > \end{align} > $ > > Because higher $\omega$ are staying at one or decreasing and lower $\omega$ gets smaller, this is a [[High Pass Filter]]. > > A similar [[Low Pass Filter]] can be obtained by letting $a_{1} <0$ ie. $a_{1}\in[-1,0]$ but closer to $-1$. $\huge \begin{align} y_{t} &= x_{t}-x_{t-1}+x_{t-2} \\ \mathcal{H}(z)&= 1-z^{-1}+z^{-2} \\ H(z) &=1-e^{-i\omega}+e^{-i 2\omega} \end{align} $ This case is palindromic hence has a [[Linear Phase|linear phase function]]. $\huge \begin{align} H(\omega) &= e^{-i \omega} \left( e^{i \omega}-1+e^{-i \omega} \right) \\ &=e^{-i \omega} \left( 2\cos \omega - 1\right) \\ \mathcal{H}(z) &= (1-z^{-1}+z^{-2}) \frac{z^{2}}{z^{2}} \\ &= \frac{z^{2}-z+1}{z^{2}} \\ H(\omega) &=\frac{e^{i {2} \omega } - e^{i \omega}+1}{e^{i 2 \omega}} \\ |H(\omega)| &= \left| e^{i 2 \omega} - e^{i \omega} + 1 \right| \\ \end{align} $ $\huge \begin{align} \mathcal{H}(z) &= \left( z - \left( \frac{1}{2} + i \frac{\sqrt{ 3 }}{2} \right) \right) \left( z - \left( \frac{1}{2} - i \frac{\sqrt{ 3 }}{2} \right) \right) \\ &=\left( z-e^{i \frac{\pi}{3}} \right)\left( z- e^{-i \frac{\pi}{3}} \right) \\ |H(\omega)| &= |e^{i\omega}-e^{i \frac{\pi}{3}}| \cdot|e^{i \omega}-e^{-i \frac{\pi}{3}}| \end{align} $ This shows that this filter decreases the amplitudes of frequencies close to $\pm \frac{\pi}{3}$. ## Filters with Phasors How we get $H(\omega)$![[Frequency Response .excalidraw.svg]] by inserting $e^{i\omega t}$ into the filter as $x_{t}$, and get $y_{t}$ as $H(\omega) e^{i\omega t}$. ie. $\huge \begin{align} y_t &=e^{i\omega t}+e^{i \omega(t-1)} \\ &= e^{i \omega t} \underbrace{\left( 1+e^{-i\omega} \right)}_{H(\omega)} \end{align} $ So the frequency response of a [[Signal Filter|filter]] is the [[Complex Numbers|Complex Number]], algebraically dependent of $\omega$, which multiples $e^{i \omega t}$ when this [[Phasor]] is input into the [[Signal Filter|filter]].