loadind data loop fastly?

Question

denis bertin 2017-10-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/360523-loadind-data-loop-fastly

评论： OCDER 2017-10-11

Hi everybody, Please i have this code that load data too long time, i want to read it fastly(reduce time execution), somebody can i help me?

%Data is the 381x247x3 matrix; if true

    LastGood = floor((122.5 - 50) / Header.Fstep) + 1 - 1; % value =29
    FirstGood = ceil((140 - 50) / Header.Fstep) + 1 + 1;  % value =38
    disp(LastGood); % value =29
    disp(FirstGood); % value =38
    disp(size(Data, 2)); % value =247
    for(k = 1:size(Data, 2))
        InterpolazioneReale1 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,real(Data(1:LastGood,k,1)),real(Data(FirstGood:end,k,1))),'splineinterp');
        InterpolazioneReale2 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,real(Data(1:LastGood,k,2)),real(Data(FirstGood:end,k,2))),'splineinterp');
        InterpolazioneReale3 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,real(Data(1:LastGood,k,3)),real(Data(FirstGood:end,k,3))),'splineinterp');
        InterpolazioneImmaginaria1 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,imag(Data(1:LastGood,k,1)),imag(Data(FirstGood:end,k,1))),'splineinterp');
        InterpolazioneImmaginaria2 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,imag(Data(1:LastGood,k,2)),imag(Data(FirstGood:end,k,2))),'splineinterp');
        InterpolazioneImmaginaria3 = fit(cat(2,[1:LastGood],[FirstGood:381])',cat(1,imag(Data(1:LastGood,k,3)),imag(Data(FirstGood:end,k,3))),'splineinterp');
        Data(LastGood + 1:FirstGood - 1,k,1) = InterpolazioneReale1(LastGood + 1:FirstGood - 1) + i * InterpolazioneImmaginaria1(LastGood + 1:FirstGood - 1);
        Data(LastGood + 1:FirstGood - 1,k,2) = InterpolazioneReale2(LastGood + 1:FirstGood - 1) + i * InterpolazioneImmaginaria2(LastGood + 1:FirstGood - 1);
        Data(LastGood + 1:FirstGood - 1,k,3) = InterpolazioneReale3(LastGood + 1:FirstGood - 1) + i * InterpolazioneImmaginaria3(LastGood + 1:FirstGood - 1);
    end
  end

6 个评论
显示 4更早的评论隐藏 4更早的评论

Jan 2017-10-10

编辑：Jan 2017-10-10

在 MATLAB Online 中打开

Note:

cat(2,[1:LastGood],[FirstGood:381])'

can be simplified to:

[1:LastGood, FirstGood:381]'

Then store this in a variable instead of creating it 6 times.

Performing 6 spline fits for the same locations seems rather inefficient. Could this be simplified?

denis bertin 2017-10-10

编辑：denis bertin 2017-10-10

Hi JAN SIMON, Yes, could be simplified, but i don't know how! I changed cat(2,[1:LastGood],[FirstGood:381])' to one variable [1:LastGood, FirstGood:381]', there is just a few change performance.

Thank you.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

OCDER 2017-10-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/360523-loadind-data-loop-fastly#answer_285153

编辑：OCDER 2017-10-10

在 MATLAB Online 中打开

Not sure if this is faster or not, but your code could be simplified a lot. If you have parallel computing toolbox, you could divide slow jobs across N cores.

    LastGood = floor((122.5 - 50) / Header.Fstep) + 1 - 1; % value =29
    FirstGood = ceil((140 - 50) / Header.Fstep) + 1 + 1;   % value =38
    disp(LastGood);     % value = 29
    disp(FirstGood);    % value = 38
    disp(size(Data, 2));% value = 247
    X = [1:LastGood, FirstGood:381]';
    parfor z = 1:3
        Tdata = Data(:, :, z);
        for k = 1:size(Tdata, 2) %Or do parfor here instead
            InterpReal = fit(X, real(Tdata(X, k)), 'splineinterp');
            InterpImag = fit(X, imag(Tdata(X, k)), 'splineinterp');
            Range = LastGood + 1:FirstGood - 1;
            Tdata(Range, k) = InterpReal(Range) + 1j * InterpImag(Range); %NOTE: 1j is now sqrt(-1)
        end
        Data(:, :, z) = Tdata;
    end