建立一个numpy矩阵

Question

I'm trying to build a matrix in numpy. The matrix dimensions should be (5001x7). Here is my code:

S=np.array([.0788,.0455,.0222,.0042,.0035,.0029,.0007])
#This is vector S, comprised of 7 scalars.

lamb=list(range(0,5001))
#This is a list of possible values for lambda, a parameter in my data.

M = np.empty([5001,7], order='C')
#This is the empty matrix which is to be filled in the iterations below.

for i in S:
    for j in lamb:
         np.append(M,((S[i]**2)/(lamb[j]+S[i]**2)))

The problem I'm having is that M remains a matrix of zero vectors.

Important details: 1) I've assigned the final line as:

    M=np.append(M,((S[i]**2)/(lamb[j]+S[i]**2)))

I then get an array of values of length 70,014 in a 1d array. I'm not really sure what to make of it.

2) I've already tried switching the dtype parameter between 'float' and 'int' for matrix M.

3) I receive this warning when I run the code: VisibleDeprecationWarning: using a non-integer number instead of an integer will result in an error in the future app.launch_new_instance()

4) I'm working in Python 3.4

I really appreciate your help. Thank you!

Answer 1

np.append makes a copy of the array and appends values to the end of the copy (making the array larger each time), whereas I think you want to modify M in place:

 for i in range(len(S)):
     for j in range(len(lamb)):
          M[j][i] = ((S[i]**2)/(lamb[j]+S[i]**2))

Answer 2

1) append adds to the end of the array, which is why your final array has 5001x7x2=70014 elements. Only the first half is zeros. It flattens the array to 1D because you didn't specify an axis to append.

2) A much more "numpy" way to do this whole process is broadcasting

S=np.array([.0788,.0455,.0222,.0042,.0035,.0029,.0007])
lamb=np.arange(0,5001)
M=(S[:,None]**2)/(lamb[None,:]+S[:,None]**2)