Talk:Total circles area: Difference between revisions
Content added Content deleted
(Monte Carlo) |
|||
| Line 2: | Line 2: | ||
A semi-important point and semi-nitpicking for doing Monte Carlo: you need to estimate the error in a credible way given the number of samples, and should ''not'' output more significant digits than the confidence allows. The current C code sample size of 100M can't possibly allow 8 decimal point precision. --[[User:Ledrug|Ledrug]] 04:29, 16 September 2012 (UTC) |
A semi-important point and semi-nitpicking for doing Monte Carlo: you need to estimate the error in a credible way given the number of samples, and should ''not'' output more significant digits than the confidence allows. The current C code sample size of 100M can't possibly allow 8 decimal point precision. --[[User:Ledrug|Ledrug]] 04:29, 16 September 2012 (UTC) |
||
: Right, in the C/Python outputs only the first few digits are correct. Feel free to add a bit of error estimation code. |
|||
Revision as of 10:35, 16 September 2012
Monte Carlo
A semi-important point and semi-nitpicking for doing Monte Carlo: you need to estimate the error in a credible way given the number of samples, and should not output more significant digits than the confidence allows. The current C code sample size of 100M can't possibly allow 8 decimal point precision. --Ledrug 04:29, 16 September 2012 (UTC)
- Right, in the C/Python outputs only the first few digits are correct. Feel free to add a bit of error estimation code.