since $ x(t)=e^{2jt} $ yields $ y(t)=te^{-2jt} $ and $ x(t)=e^{-2jt} $ yields $ x(t)=te^{2jt} $, it is easy to see that x(t) yields y(t)=t*x(-t).
Based on the above, $ x(t)=cos(2t) $ would yield $ y(t)=tx(-t)=tcos(-2t) $.
The above solution works for this case, but it does not show the properties of linearity.
If $ x_1(t)=e^{2jt} $ and $ x_2(t)=e^{-2jt} $, Take the input signal to be $ \frac{1}{2}(x_1(t)+x_2(t)) $. The response to the input is $ y(t)=\frac{1}{2}y_1(t)+\frac{1}{2}y_2(t) $ because the system is linear.
$ \frac{1}{2}(e^{2jt}+e^{-2jt})\rightarrow system\rightarrow t(\frac{1}{2}e^{-2jt}+\frac{1}{2}e^{2jt}) $
$ t(\frac{1}{2}e^{-2jt}+\frac{1}{2}e^{2jt})=t(\frac{1}{2}(e^{-2jt}+e^{2jt}) $
$ =t(\frac{1}{2}(cos(-2t)+jsin(-2t)+cos(2t)+jsin2(t)) $ $ =t(\frac{1}{2}(cos(2t)-jsin(2t)+cos(2t)+jsin2(t)) $ $ =t(\frac{1}{2}(2cos(2t)) $ $ =tcos(2t) $
