So, I know there are a few functions that don't have elementary antiderivatives, and the way I approach some of these problems is by integrating the Taylor series. In the process, I usually interchange the (indefinite) integral sign and the summation sign. Example: $$ \int \sin(x)/x\ dx\,=\\\int \sum_{n=0}^\infty (-1)^nx^{2n}/(2n+1)!\ dx=\\\sum_{n=0}^\infty ((-1)^{n}/(2n+1)!)\int x^{2n} dx=\\C+\sum_{n=0}^\infty (-1)^{n}x^{2n+1}/((2n+1)!(2n+1)) $$ Is this valid? when doing indefinite integrals of this sort (integrating Taylor series), when, if ever, will it be invalid to interchange the integration and summation signs?
Asked
Active
Viewed 69 times
1
-
1It's usually appropriate to make such maneuvers, as both Riemann Integration and Lesbegue Integration are consistent under such swappages. I personally find baby Rudin to be a good guide for this question, although any elementary analysis text should do the question justice. – Rushabh Mehta Aug 10 '18 at 02:15
-
1A power series can be integrated term-by-term. That is, for Taylor series you can always interchange integration and summation with the circle of convergence. – saulspatz Aug 10 '18 at 02:28
-
2Various answers here. – David Aug 10 '18 at 02:31