I ran across this peculiarity in Mathcad 11 (Maple), and wondered if any of the newer Mathcad or Prime versions have the same ..... Bug.
First. This seems normal:
But if a is other than 0, e.g. 2 then:
The 3rd integration (Phi3) should start at 0 for t=2. It doesn't. At that point even the numeric and symbolic processors disagree, with conflicting and both wrong answers.
Now the bug is easily circumvented by first generally (boundaryless) integrating the function at hand, and subsequently taking the bounded integral:
But this shouldn't be...
Luc