# Issues with symbolic integration

I am trying to do the following indefinite integral,

Integrate[(2*((2*I)*E^(I*t2) + (2*I)*E^((2*I)*y))*
PolyLog[2, 1 - E^(I*(t2 - y))])/(3*E^((I/2)*(t2 + 2*y))) +
(2*((-2*I)*E^(I*t2) - (2*I)*E^((2*I)*y))*
PolyLog[2, 1 - E^((-I)*y)])/(3*E^((I/2)*(t2 + 2*y))), y]


And it doesn't give me any result, but on breaking it term by term and integrate it works! What is going on ?

• Possible explanation. Mathematica tries certain transformations for a limited time before giving up. Splitting an integral into simpler parts may give it time to find a solution. – mikado May 15 '19 at 5:43
• @mikado : So can we change the settings to tell mathematica to be more persistent ? Or is there an issue with my Macbook Air as it is a sub-performing machine ? – Jaswin May 15 '19 at 8:42
• I think my previous suggestion is wrong, as Mathematica gives up trying almost immediately. Another possibility relates to branch cuts. For indefinite integrals, Mathematica returns an answer that is correct for an assumed path. If we split the integral, it can use a different path for each component. Combined, it may not be able to find a suitable path. – mikado May 15 '19 at 19:45