Timeline for Increase performance of Bayesian likelihood equation
Current License: CC BY-SA 3.0
8 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Mar 18, 2017 at 17:05 | comment | added | Gabriel |
Here's a nice tutorial on numpy's broadcasting.
|
|
| Mar 18, 2017 at 16:50 | comment | added | hpaulj |
More about increasing performance with numpy: codereview.stackexchange.com/questions/17702/…
|
|
| Mar 18, 2017 at 16:48 | comment | added | hpaulj |
Because it replaces Python loops with compiled numpy code. Sorry. I normally answer on SO where this kind of numpy 'vectorization' is an everyday question.
|
|
| Mar 18, 2017 at 16:41 | comment | added | Vogel612 | okay ... nice that this is faster, but ... WHY??? | |
| Mar 18, 2017 at 9:59 | comment | added | Graipher |
You can get another few percent faster by replacing sum(np.log(sum_N)) with np.log(sum_N).sum() or np.sum(np.log(sum_N)).
|
|
| Mar 17, 2017 at 21:54 | vote | accept | Gabriel | ||
| Mar 17, 2017 at 21:54 | comment | added | Gabriel | Excellent! This code is between 80-100 times faster in my machine, thank you very much hpaulj! | |
| Mar 17, 2017 at 21:45 | history | answered | hpaulj | CC BY-SA 3.0 |