Linspace to logspace smoothing

Question

Marco Clementi 2023-7-14

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1995833-linspace-to-logspace-smoothing

回答： Supraja 2023-7-27

Hi, I want to interpolate some linearly spaced data to a logarithmic grid for plotting.

The first solution I tried was:

function [f_log,data_log] = lin2log(f_lin,data_lin)
    NPOINTS=1001;
    f_lin=log10(f_lin);
    f_log=log10(logspace(f_lin(1),f_lin(end),NPOINTS));
    data_log=interp1(f_lin,data_lin,f_log);
    f_new=10.^f_new;
end

However, the input data is very noisy, and I would like to also smooth it before interpolating. So I updated the code with a for loop:

function [f_log,data_log] = lin2log(f_lin,data_lin)
    NPOINTS=1001;
    f_lin=log10(f_lin);
    f_log=log10(logspace(f_lin(1),f_lin(end),NPOINTS));
    for nn=1:NPOINTS-1
        inds = (f_lin>=f_log(nn) & f_lin<(f_log(nn+1)));
        data_lin(inds)=mean(data_lin(inds));
    end
    data_log=interp1(f_lin,data_lin,f_log);
    f_new=10.^f_new;
end

This worked nicely, but the loop seems to slow down the code quite a lot, especially given that the input data is a very long array.

Any suggestions on a way improve the code efficiency? Many thanks!

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

KSSV 2023-7-14

Did you try functions like smooth

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Supraja 2023-7-27

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1995833-linspace-to-logspace-smoothing#answer_1279777

You can improve you code efficiency by sorting the data and then passing it to the “interp1” function.

You can also use “smooth” function for better efficiency.

The detailed documentation link of the “interp1” function is given here:https://www.mathworks.com/help/matlab/ref/interp1.html#btwp6lt-3

You can also refer to the MATLAB Answer given here which is similar to your query:https://www.mathworks.com/support/search.html/answers/113651-how-to-smoothing-a-curve.html?fq%5B%5D=asset_type_name:answer&fq%5B%5D=category:curvefit/smoothing&page=1

Hope this helps!