The filtfilt function is different in different versions

When I used filtfilt in version R2019b, I had to use input data in double format, but in version R2024, functions can recognize data in single format.But when I increased the order of the filter from second to third order, it became clear that the single format version produced the wrong result, and the result using the double version is still right.

I want to know why, please help me.

[Infinite_king]

Can you provide the exact input values ?

[Esther]

The filtfilt function support single-precision data since R2023a:

https://ww2.mathworks.cn/help/signal/release-notes.html?rntext=filtfilt&startrelease=R2021a&endrelease=R2024a&groupby=release&sortby=descending&searchHighlight=filtfilt#mw_ac2738e0-95bf-4fb1-9676-5f5038405253

Can you share your code to reproduce the problem in 24a?

Answer 1

[Abhiram]

Hi @中祥 范

When single-precision input is given, the filter coefficients of the ‘filtfilt’ function will be quantized to single precision.

The incorrect results with single precision data are especially evident when the filter order increases, as the numerical precision required exceeds what single precision can reliably provide.

Use of double-precision input can offer better stability and accuracy, which is crucial for higher-order filters.