If integral which contains an oscillating function, e.g. a Bessel function, is solved with a trapezoidal rule I am getting that the answer of the integral is fluctuating about value when I increase the sample size and continuously increasing the sample size increasingly converges the answer to this value. Is this behavior reasonable for a fluctuating function as the integrand?
Here is a picture to make the question a bit clear (please ignore the CLASS curve in the figure below)
