Converting Python to Matlab

Spent a few hours on trying to figure out why the outputs are different, no luck. Python and Matlab are in a txt file along with their outputs.

Suggestions on what I should be looking at to resolve the issue?

[Walter Roberson]

We do not have the data or NumPanels to test with.

[Zach Dunagan]

I figured it out.

Answer 1

[Andrei Bobrov]

Small typo in MATLAB code in nu2

    nu2 = atan2(-(xc(k)-xp(n))*sin(theta(k))+(yc(k)-yp(n))*cos(theta(n))+(xp(n+1)-xp(n))*sin(theta(n))-(yp(n+1)-yp(n))*cos(theta(n)),(xc(k)-xp(n))*cos(theta(n))+(yc(k)-yp(n))*sin(theta(n))-(xp(n+1)-xp(n))*...
  cos(theta(n))-(yp(n+1)-yp(n))*sin(theta(n))); % fixed!

[Zach Dunagan]

I figured it out. I don't need the multiplication, '\' in Python means '...' in Matlab. The nu2 = atan2(x1, x2) similar to this line of code in my script. atan2(yp(k+1)-yp(k), xp(k+1)-xp(k)).

[Andrei Bobrov]

Stop Zach!

How?

cos(theta(n)-(yp(n+1)-yp(n))*sin(theta(n)))

or

cos(theta(n))-(yp(n+1)-yp(n))*sin(theta(n))

[Zach Dunagan]

Oh... Thanks.

Answer 2

[Zach Dunagan]

How would I insert zeros in the diagonal line of these 128 x 128 matrices?

nSource = -(sin(theta-transpose(theta)))*ups + (cos(theta-transpose(theta)))*wps;  
ftSource = cos(theta-transpose(theta))*ups + sin(theta-transpose(theta))*wps;
nVortex = sin(theta-transpose(theta))*upv + cos(theta-transpose(theta))*wpv;
tVortex = cos(theta-transpose(theta))*upv + sin(theta-transpose(theta))*wpv;

[Walter Roberson]

For the special case of square matrices, often the easiest way is

M - diag(diag(M))

For other matrices,

 tril(M,1) + triu(M,1)

[Zach Dunagan]

What if I wanted to insert a number, other than zero?

[Walter Roberson]

Just in case you have unsigned integers or the case where the existing value is so different from the new value that you have to worry about loss of precision:

Square matrix:

M(1:size(M,1)+1:end) = NewNumber;

Non-square matrix:

shorter = min(size(M,1),size(M,2));
lastidx = size(M,1)*(shorter-1)+shorter;
M(1:size(M,1)+1:lastidx) = NewNumber;

[Zach Dunagan]

M = M = nSource = 128 x 128 ?

Is the M inside of the size() the vector that lays in the diag of nSource?

[Walter Roberson]

M is the array whose diagonal is to be set. For example,

nSource(1:size(nSource,1)+1:end) = NewNumber;

if nSource is square.

[Zach Dunagan]

Actually this seem to have worked for my situation.

ups(logical(eye(size(ups)))) = 0;

wps(logical(eye(size(wps)))) = 0.5;

[Zach Dunagan]

Can you please help me with the Matlab equivalent?

A=np.zeros((numPanels+1,numPanels+1))

A[:numPanels,:numPanels]=nSource

I just need the equivalent to this code.

A[:numPanels,-1]=np.sum(nVortex,axis=1)

[Walter Roberson]

A[:numPanels,:numPanels]=nSource

would be

A(1:end-1, 1:end-1) = nSource;

I am not certain about

A[:numPanels,-1]=np.sum(nVortex,axis=1)

but I suspect

 A(1:end-1, end) = sum(nVortex, 2);

[Zach Dunagan]

It worked! Thank you.

[Zach Dunagan]

A[-1,:numPanels]=tSource[0,:]+tSource[-1,:]

Here is what I have.

A(end, 1:end) = tSource(1, :) + tSource(end, :);

Error:

Subscripted assignment dimension mismatch.

Error in ssPanelMethod (line 112) A(end, 1:end) = tSource(1, :) + tSource(end, :);

[Andrei Bobrov]

A(end, 1:end-1) = tSource(1, :) + tSource(end, :);

[Zach Dunagan]

Thank you.

[Zach Dunagan]

I really want to understand these little operations, rather than constantly asking for help here...

Python code

normVelFoil=np.dot(nSource,x[:-1])+np.dot(nVortex,x[-1]*np.ones((num
Panels,1)))+normU

tangVelFoil=np.dot(tSource,x[:-1])+np.dot(tVortex,x[-1]*np.ones((numPanels,1)))+tangU

Matlab code

normVelFoil = dot(nSource, x(1:end, 1:end-1)) + dot(nVortex, x(1:end, 1:end-1) * ones(numPanels, 1)) + normU;

tangVelFoil = dot(tSource, x(1:end, 1:end-1)) + dot(tVortex, x(1:emd, 1:end-1) * ones(numPanels, 1)) + tangU;

tSource, tVortex are 128 x 128, while x is a 129 x 1.

[Zach Dunagan]

I changed dot to mtimes() or I could use .*

I compare the first dot to second dot of python and everything looks to be lining up. When I compare the normVelFoil outputs they don't line up.

Here is what I changed..

normVelFoil = mtimes(nSource, x(1:end-1, 1)) + mtimes(nVortex, x(end)*ones(numPanels, 1)) + normU;

[Zach Dunagan]

I have come to a conclusion the outputs are the same. They are both extremely small values, practically zero.

Answer 3

[ASHOK KUMAR MEENA]

def Lagrange(x, y, n, xx):

sum = 0

for i in range(0, n + 1):

product = y[i]

for j in range(0, n + 1):

if (i != j):

product = product * (xx - x[j]) / (x[i] - x[j])

sum += product

return sum

def Trapezoidal(h, n, f):

sum = f[0]

for i in range (1, n):

sum = sum + 2 * f[i]

sum = sum + f[n]

ans = h * sum / 2

return ans