Interesting differentiation regarding $\delta$-functions. Let we define a function $h(x)=e^{-ax}H(x)+e^{ax}H(-x)$ and we want to find the $n$-th derivative. \begin{align} h(x)&=e^{-ax}H(x)+e^{ax}H(-x)\\ h'(x)&=e^{-ax}\delta(x)-ae^{-ax}H(x)-e^{ax}\delta(x)+ae^{ax}H(-x)=ae^{ax}H(-x)-ae^{-ax}H(x)\\ \end{align} I believe this step has no problem, but the following difference arises. If we start from the resulting expression of the first derivative, then \begin{align} h''(x)&=(h'(x))'=-e^{ax}a\delta(x)+a^2e^{ax}H(-x)-e^{-ax}a\delta(x)+a^2e^{-ax}H(x)=a^2e^{ax}H(-x)+a^2e^{-ax}H(x)-(e^{-ax}+e^{ax})a\delta(x) \end{align} but if we start before we discard $(e^{-ax}-e^{ax})\delta(x)=0$ which has the derivative $-a(e^{-ax}+e^{ax})\delta(x)+(e^{-ax}-e^{ax})\delta'(x)$, then \begin{align} h''(x)&=-a(e^{-ax}+e^{ax})\delta(x)-e^{ax}a\delta(x)+a^2e^{ax}H(-x)-e^{-ax}a\delta(x)+a^2e^{-ax}H(x)=a^2e^{ax}H(-x)+a^2e^{-ax}H(x)-(e^{-ax}+e^{ax})a\delta(x)\\ &=a^2e^{ax}H(-x)+a^2e^{-ax}H(x)-2ae^{-ax}\delta(x). \end{align}
Though the results are equal due to $\delta$'s property, but what are the strictly correct way to proceed further.
Could be any easier way to arrive at the $n$-derivatives?
If $h$ is a distribution given by
$$h(x)=e^{-ax}H(x)+e^{ax}H(-x)$$
then the distributional derivative, $h'$, of $h$ is given by
$$h'(x)=\underbrace{e^{-ax}\delta(x)-e^{ax}\delta(x)}_{=0}+ae^{ax}H(-x)-ae^{-ax}H(x)\tag1$$
Now, if we differentiate the first two terms on the right-hand side of $(1)$ we find that
$$\begin{align} 0&=(e^{-ax}\delta(x)-e^{ax}\delta(x))'\\\\ &=(-2\sinh(ax)\delta(x))'\\\\ &=-2\sinh(ax)\delta'(x)-2a\cosh(ax)\delta(x)\\\\ &=2a\cosh(ax)\delta(x)-2a\cosh(ax)\delta(x)\\\\ &=0 \end{align}$$
as expected!