On Thu, Oct 11, 2018 at 9:32 AM David Thornton via talk <[email protected]> wrote: > > This is a chance for me to test my understanding by trying to explain it > myself. > > Monte Carlo is where you randomly choose solutions from a solution space and > then try to make the results better for each choice. > > Imagine you are trying to find the highest point on earth. Rather than scan > the entire earth to find the highest point, you choose a sample of points and > then see if there is something higher near-by. > > This is _in general_ more efficient than scanning the whole earth, but of > course there is a change that you miss Everest. > > Consider the alternative approach of choosing a point and then looking for > higher points nearby ( aka Gradient Descent). Suppose you started somewhere > in Britain... you might scan the local area and come to the conclusion that > snowdon is the highest place in the world. > > Now image that you do this 100 time, and then compare the local bests of > those 100 attempts. You are more likely to find the highest point. That's > monte carlo I think.
Its a little different than that but you're headed in the right direction. The process was developed as the first atomic bomb was being developed in the USA (fascinating as a large number of those scientists were expat Germans (early 1940s to 1945). > > Anyway, I'm not sure this has anything to do with what the OP is talking > about. > > If I understand he/she wants to recreate an app. Does that mean reverse > engineer? Are you trying to recreate someone else's app without direct access > to that app? > > Are you trying to examine interactions with the app and predict what the app > will do, and hence "simulate" the app? > > You might also consider creating a _smart_ proxy, that lets you cache results > locally such that if you ask for the same thing a second time it's comes back > faster. > > Additionally by implementing a proxy , you can populate your own copy of what > ever data the original service is providing and study that for the purposes > of reverse engineering. > > This sounds like a great conversation to have over a $drink, and/or in front > of a white board. I like your last idea except I would have to travel about 2500 km and that produces its own 'fun'. Thank you for your ideas/assistance and if you know an easy way to communicate 'chat at remote' - - - well I'm listening. Dee --- Talk Mailing List [email protected] https://gtalug.org/mailman/listinfo/talk
