I've been trying to use gsl_integration_qags(&F, 0, 20, 0, 1e-2, 1000, w, &result, &error) to integrate a function defined by a large set of sample values taken at a constant rate.
I provided a routine for F which interpolates (linearly) between sample values. I've tried to play with epsabs and epsrel but get either: "cannot reach tolerance because of roundoff error" or "integral is divergent, or slowly convergent". BTW, simply using Simpson's rule I get: 0.148525. The sample values range from 3.5320 and -1.0966 with delta-t being a constant 0.003125. Any ideas? ---John
