So I have some code simulating the behavior of a complex system. The system works perfectly (with a "fidelity" of $f=1$) for some parameter $\beta=0$. For other values of $\beta$ the fidelity will be some number between 0 and 1. I know that $f(\beta)$ peaks for $\beta=0$ (not sure how sharply exactly) and then drops off to some kind of noisy floor for larger absolute values of $\beta$. I don't have any analytic insight into $f(\beta)$ as a function.

Now, if I draw $N$ values $\beta$ from a Gaussian distribution with mean 0 and a fixed standard deviation $\sigma_\beta$, and then calculate the mean fidelity as $\bar f(\sigma_\beta) = \frac{1}{N}\sum_{n=1}^N f(\beta_n)$, is there a way for me to estimate the standard error on $\bar f$? Or, alternatively, is there a good way for me to figure out how many samples $N$ I need to take to get a value of $\bar f$ that I feel confident about within 1%?

This might be a fairly basic question, but it's been a long time since my undergraduate days ;-). In the absence of anything quantitative, I'd probably just try pulling 100 samples and check how stable the resulting $\bar f$ is compared to if I just used 50 samples (i.e., throw out every other sample).



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.