從MATLAB轉換的代碼段爲C＃

我試圖從matlab此功能轉換爲C#從MATLAB轉換的代碼段爲C＃

什麼是這些matlab語句相當於C#？下面的C#部分顯示的matlab代碼部分，我不知道如何移植到C#

res(res == 0) = p * scale * unit(res == 0); 
    W = min(unit, scale * p * (abs(res)).^ (-1));

整個matlab部分，我試圖將它移植

function W = huber(res, scale, param) 
% function W = huber(res, scale, param) 
% 
% computes Huber's weight function for robust regression: 
% min(1, param/(|res|/scale)) 
% 
% arguments: 
%  res: vector of residuals 
%  scale: robust estimate of scale, such as MAD 
%  param: parameter of the Huber function; the 
%   default value is 2.5. 
% 
% returns: 
%  W: the vector of Huber weights 
% 
% 
% P.B. Stark [email protected] 
% 9 July 1997. 

p = 2.5;  % default parameter 
if (nargin == 3), 
    if (param > 0), 
     p = param; 
    else 
     error('parameter must be positive') 
    end 
end 

unit = ones(size(res)); 
res(res == 0) = p*scale*unit(res==0); 
W = min(unit, scale*p*(abs(res)).^(-1)); 
return;

C＃代碼。

public static double Hueber(double[] residuals, double scale, double param) 
{ 
     //Calculation of the mean estimate for a given range using Huber's weights 
     //The Huber's function is defined as min(1, param/(|residuals|/scale)) 
     //param: a given parameter - affects the range where weights = 1 
     //residuals: deviation from the mean estimate (I used the median as the first approximation) 
     //scale: estimate of variation. Some use 1.483*(median absolute deviation, or MAD, 
     // of the deviations of the data from their median). I used std, i.e. dblAveVar 
     //See http://www.stat.berkeley.edu/~stark/Preprints/Oersted/writeup.htm 

     double p = 2.5;  //default parameter 
     if (param > 0) 
      p = param; 
     else 
      throw new Exception("parameter must be positive"); 

     double[] unit = Ones(residuals.Length); 

     // What is the C# of this? 
     //res(res == 0) = p * scale * unit(res == 0); 
     //W = min(unit, scale * p * (abs(res)).^ (-1)); 

     return W; 
    } 

    public static double[] Ones(int length) 
    { 
     double[] ones = new double[length]; 

     ones.Populate(1.0); 

     return ones; 
    }

來源

2017-09-15 Ivan

如果您說出代碼應該做什麼會更容易。 – HimBromBeere

Btw你的默認參數值沒有用處。 'p'的值總是被'param'覆蓋或者拋出一個異常，而原始代碼中沒有這樣的情況，它具有基於'nargin'的值的額外檢查（這裏不存在） – pinkfloydx33

從複習matlab docs'res（res == 0）= X'表示將res的索引值爲零的res的所有索引設置爲X.因此，通過res進行迭代，如果值爲零，則將值替換爲公式在右邊。單元（res == 0）表示res中相應索引值爲零的單元索引。 **我認爲**。除了快速查看他們的文檔外，我對Matlab一無所知。你必須知道matlab如何處理乘法向量並獲取絕對值。最終，我認爲這將是一個非常迭代的解決方案。 – pinkfloydx33

//res(res == 0) = p * scale * unit(res == 0); 
//W = min(unit, scale * p * (abs(res)).^ (-1));

代碼在下面給出;

public static double Hueber(double[] residuals, double scale, double param) 
{ 
    double p = 2.5;  //default parameter 
    if (param > 0) 
    p = param; 
    else 
    throw new Exception("parameter must be positive"); 

    for (int i = 0; i < residuals.Length ; i++) { 
    if (residuals[i] == 0) { 
     residuals[i] = p * scale * 1; // res(res == 0) = p*scale*unit(res==0) 
    } 
    // Then we can do smae step in this loop W = min(unit, scale*p*(abs(res)).^(-1)) 
    residuals[i] = Math.Min(1, scale*p*Math.Abs(1/residuals[i])); 
    } 

    return residuals; // w = residuals[] 
    }

來源

2017-09-15 22:20:55

我編輯完整的代碼，你不需要一個功能....有一個很好的編碼:) –

從MATLAB轉換的代碼段爲C＃

回答

相關問題