Help me with indexing please?

Question

Zach Dunagan 2017-10-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/362452-help-me-with-indexing-please

编辑： Stephen23 2017-10-22

Trying to convert Python to Matlab. Below is the Python code along with the output.

import numpy as np
import matplotlib.pyplot as plt
import copy
# setup panel coordinates and control point coordinates
AoA=10*np.pi/180
Uinf=8.8
rho=1.204
chord=0.12
span=chord*2
thickness=0.006
minRad=thickness/2
majRad=chord/4
numPanels=128
xp=np.zeros((numPanels+1,1))
yp=np.zeros((numPanels+1,1))
# panel coordinates for a finite thickness plate with elliptical leading edges (nodes)
xStep=chord/(numPanels/2)
xp[0]=chord/2
yp[0]=0
i=1
while i<numPanels/8:
    xp[i]=chord/2-xStep*i
    yp[i]=-minRad/majRad*(majRad**2-(majRad-xStep*i)**2)**(1/2)
    i=i+1
xp = 
0.06
0.058125
0.05625
0.054375
0.0525
0.050625
0.04875
0.046875
0.045
0.043125
0.04125
0.039375
0.0375
0.035625
0.03375
0.031875
yp = 
0
-0.00104396
-0.00145237
-0.00174888
-0.00198431
-0.00217855
-0.00234187
-0.00248039
-0.00259808
-0.00269766
-0.00278107
-0.00284975
-0.00290474
-0.00294679
-0.00297647
-0.00299413

Here is my Matlab Code with the output.

    % clears variables and windows
    clc, clear all
    % user defined variables
    AoA = 10*pi/180;
    Uinf = 8.8;
    rho = 1.204;
    chord = 0.12;
    span = chord*2;
    thickness = 0.006;
    minRad = thickness/2;
    majRad = chord/4;
    numPanels = 128;
    xp = zeros(numPanels, 1);
    yp = zeros(numPanels, 1);
    % panel coordinates for a finite thickness plate with elliptical leading
    % edge (nodes)
    xStep = chord/(numPanels/2);
    k = 1;
    while k < numPanels/9
       xp(k) = chord/2*k;
       yp(k) = 0*k;
       xp(k) = chord/2 - xStep*k;
       yp(k) = -minRad/majRad*(majRad^2-(majRad-xStep*k)^2)^(1/2);
       k = k + 1;
    end
    xp =
    0.0581250000000000
    0.0562500000000000
    0.0543750000000000
    0.0525000000000000
    0.0506250000000000
    0.0487500000000000
    0.0468750000000000
    0.0450000000000000
    0.0431250000000000
    0.0412500000000000
    0.0393750000000000
    0.0375000000000000
    0.0356250000000000
    0.0337500000000000
    yp = 
    -0.00104395581803063
    -0.00145236875482778
    -0.00174888357245415
    -0.00198431348329844
    -0.00217855313224167
    -0.00234187424939940
    -0.00248039185412305
    -0.00259807621135332
    -0.00269765523186341
    -0.00278107443266087
    -0.00284975327879450
    -0.00290473750965556
    -0.00294679380853157
    -0.00297647022494767

I'd like the output of Matlab to match the output of Python.

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

Stephen23 2017-10-21

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/362452-help-me-with-indexing-please#answer_286966

编辑：Stephen23 2017-10-21

在 MATLAB Online 中打开

There is no point writing MATLAB code as if it were Python (or C++) with lots of loops. MATLAB is a high-level language, so forget about those ugly loops and learn how to write neat and simple vectorized code:

AoA   = 10*pi/180;
Uinf  = 8.8;
rho   = 1.204;
chord = 0.12;
span  = chord*2;
thickness = 0.006;
minRad    = thickness/2;
majRad    = chord/4;
numPanels = 128;
% calculate:
xStep = chord/(numPanels/2);
vec   = xStep*(0:numPanels/8-1);
xp = chord/2 - vec;
yp = -minRad/majRad * sqrt(majRad^2-(majRad-vec).^2);

and checking that it gives the same output as the Python code:

>> xp(:)
ans =
                0.06
            0.058125
             0.05625
            0.054375
              0.0525
            0.050625
             0.04875
            0.046875
               0.045
            0.043125
             0.04125
            0.039375
              0.0375
            0.035625
             0.03375
            0.031875
>> yp(:)
ans =
                   0
-0.00104395581803063
-0.00145236875482778
-0.00174888357245415
-0.00198431348329844
-0.00217855313224167
 -0.0023418742493994
-0.00248039185412305
-0.00259807621135332
-0.00269765523186341
-0.00278107443266087
 -0.0028497532787945
-0.00290473750965556
-0.00294679380853157
-0.00297647022494766
-0.00299413489175087

By the way, one of the main reasons your code does not work is because you did not take into account the differences between zero-based and one-based indexing when you calculated with the index.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Stephen23 2017-10-21

编辑：Stephen23 2017-10-22

@Zach Dunagan: well that is easy: look at every location that you use the indexing (this is easy in the MATLAB editor because of the variable highlighting), and you will realize that some of them need to have 1 subtracted from them (hint: everywhere that you multiply the index with something).

But MATLAB has quite different best-practices than Python. Using a loop is not good use of MATLAB for solving this task, because vectorized code is much simpler, is more efficient, and avoids the whole issue of indexing entirely. Also it is best to avoid while loops if the number of iterations can be clearly defined before the loop: in your case a for loop would be a more standard use of MATLAB.

If you really want to learn how to write MATLAB code then forget the loop. Note that my answer, simpler and neater without any loop, already gives the answer that you are supposedly looking for.

Zach Dunagan 2017-10-21

Thank you for your suggestions.

请先登录，再进行评论。

Answer 2

KL 2017-10-21

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/362452-help-me-with-indexing-please#answer_286967

编辑：KL 2017-10-21

在 MATLAB Online 中打开

Change the part after you define xStep with the following code

      xp_0 = chord/2;
      yp_0 = 0;
      k = 1;
      while k < numPanels/9
         xp(k) = chord/2 - xStep*k;
         yp(k) = -minRad/majRad*(majRad^2-(majRad-xStep*k)^2)^(1/2);
         k = k + 1;
      end
      xp = [xp_0; xp]
      yp = [yp_0; yp]

I have just created xp_0 and yp_0 before the loop, ran the loop from k=1 with just 2 equations (just like in python) and finally put the _0 values on top of the resultant arrays.

and the better way is to avoid the loop,

xp_0 = chord/2;
yp_0 = 0;
k = 1:floor(numPanels/9);
xp(k) = chord/2 - xStep.*k;
yp(k) = -minRad/majRad*sqrt(majRad^2-(majRad-xStep.*k).^2);
xp = [xp_0; xp]
yp = [yp_0; yp]