This method has 2 error terms - phase error and amplitude error. You can only represent sinusoids with a resolution of 2pi/2^N rad/samp (same phase error if you

This method has 2 error terms - phase error and amplitude error. You can only represent sinusoids with a resolution of 2pi/2^N rad/samp (same phase error if you don’t do phase truncation). Maybe that’s enough for most applications, but it’s probably significantly more error than math libraries that calculate cos. Totally depends on your application.