A very peaked kernel is the exponential one, whose generic expression is as follows:
Contrary to the Gaussian one, this kernel has very heavy tails and a sharp peak. A plot is shown in the following screenshot:
As it's possible to see, such a function is suitable to model very irregular distributions with a density highly concentrated around some specific points. On the other hand, the error can become very high when the data-generating process is quite regular with smooth surfaces. A good theoretical measure ...