Mex: useful comparing float values?

Question

mick strife 2013-5-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/74596-mex-useful-comparing-float-values

Hello,

i have a mex-script which returns to matlab some float-values. As i have to check the returning data i have also a c-script with the same function as the mex-script. When i want to compare the data i noticed that the matlab-data and c-data have different number of decimal places although both are using float(single) as datatype.

As i am quite a beginner in c and matlab i am not sure how to solve this usefully. thx for any advise! :)

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2013-5-4

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/74596-mex-useful-comparing-float-values#answer_84385

You have to check if they are within some tolerance of each other. Please see the FAQ. http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F . It gives you some code you can use.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Image Analyst 2013-5-4

I think c can show to more than 6 or 7 decimal places. I think you misunderstood. Maybe what they were saying was that single precision numbers are accurate only out to the 6 or 7th decimal place, and that if you show anything after that, it could be just garbage. Those are two different things to say.

Yes, I think using a tolerance, like they do in the FAQ I referred you to, is a good solution.

mick strife 2013-5-5

Thx "image analyst". Yes, i misunderstood that. He told me that float is only accurate with 6,7 decimal places and if i want more decimal places i should use an other data type.

thx for your advice! :-)

请先登录，再进行评论。