<div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr"><div dir="ltr">On Wed, Jan 30, 2019 at 10:48 AM Ilhan Polat <<a href="mailto:ilhanpolat@gmail.com">ilhanpolat@gmail.com</a>> wrote:<br></div><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div dir="ltr"><div>I am a frequent user of this function. I am also occasionally teaching control theory and I use this also professionally, I provided a proper use case as a "user". You already admitted that you don't use it. So excuse me when I say that I have more experience than you about this function API (please read this twice; function API not the underlying theory). How can I even provide evidence on this completely subjective matter? Here is all the issues related to curve_fit <br></div><div><br></div><div><a href="https://github.com/scipy/scipy/search?q=curve_fit&type=Issues" target="_blank">https://github.com/scipy/scipy/search?q=curve_fit&type=Issues</a></div><div><br></div><div>As far as I know there was no complaint so far.</div></div></div></blockquote><div><br></div><div>StackOverflow is likely a better place to look for evidence:</div><div><br></div><div><a href="https://stackoverflow.com/questions/27097957/my-use-of-scipy-curve-fit-does-not-seem-to-work-well">https://stackoverflow.com/questions/27097957/my-use-of-scipy-curve-fit-does-not-seem-to-work-well</a><br></div><div><a href="https://stackoverflow.com/questions/23828226/scipy-curve-fit-does-not-seem-to-change-the-initial-parameters">https://stackoverflow.com/questions/23828226/scipy-curve-fit-does-not-seem-to-change-the-initial-parameters</a><br></div><div><a href="https://stackoverflow.com/questions/53509550/error-in-scipy-curve-fit-for-more-than-two-parameters">https://stackoverflow.com/questions/53509550/error-in-scipy-curve-fit-for-more-than-two-parameters</a></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div dir="ltr"><div> Hence that means it is really not that big of a deal that would grant such tone. I don't know what to add other than what I already provided. You cannot make educated guesses about initial points on nonconvex searches. As you mentioned, we are as blind as np.ones(n) choice.</div></div></div></blockquote><div><br></div><div>That is certainly not the case. It is impossible to do so generally for every arbitrary nonconvex problem, which is why we can't automate it, but there are tons of problems where one has a reasonable idea from priors, examining the plotted data, or heuristic algorithms (e.g. finding the peak location and amplitude with a peak finder).</div></div><div><br></div>-- <br><div dir="ltr" class="gmail_signature">Robert Kern</div></div></div></div></div>