difficulty in understanding numpy array behavior

Question

the following is a (very)shorterned vesrion of my program

the normvecs vector is not getting overwritten , you woud be able to see for yourself just once you run this code snippet , the parameters are correct geom function takes an n+1 X 2 array and n as input, i guess i'm doing something really stupid (which i think i am) or i don't understand this behavior

import numpy as np

#geometry calculations
def geom(pts,n):
    r = np.zeros(n)
    normvecs = np.zeros((n,2))
    tgtvecs  = np.zeros((n,2))
    alphap = np.zeros(n)
    cpts = np.zeros((n,2))
    #collocation points
    cpts[:,0] = (pts[0:n,0]+pts[1:n+1,0])/2  
    cpts[:,1] = (pts[0:n,1]+pts[1:n+1,1])/2
    #length of panels
    r[:] = np.sqrt((pts[0:n,0]-pts[1:n+1,0])**2 + (pts[0:n,1]-pts[1:n+1,1])**2)
    #angle of each panel with the horizantal(chord)/refernce axis
    alphap[:] = np.arctan2(pts[1:n+1,1]-pts[0:n,1],pts[1:n+1,0]-pts[0:n,0])
    #normal vectors
    normvecs[:,0] = -np.sin(alphap[:])
    normvecs[:,1] =  np.cos(alphap[:])
    xx = np.cos(alphap[:])
    print(normvecs[:,0],normvecs[:,1],np.cos(alphap[:]),xx)
    return

geom(np.random.rand(31,2),30)

normvecs[:,1] is showing just negative of nomvecs[:,0] i wasn't able to overwrite like normvecs[:,1] = xx (no errors .. but just printed the same thing)

I maynot sound clear,once if you run you may see what i'm trying to tell.

what's the problem?(if it is so!)

talonmies · Accepted Answer · 2011-08-17 12:02:32Z

2

I see no problem at all. Running a more concise version of your code:

import numpy as np

#geometry calculations
def geom(pts,n):
    normvecs = np.zeros((n,2))
    alphap = np.zeros(n)
    alphap[:] = np.arctan2(pts[1:n+1,1]-pts[0:n,1],pts[1:n+1,0]-pts[0:n,0])
    normvecs[:,0] = -np.sin(alphap[:])
    normvecs[:,1] =  np.cos(alphap[:])
    print np.abs(normvecs[:,0])-np.abs(normvecs[:,1])
    return

geom(np.random.rand(31,2),30)

produces something like this:

[ 0.47500019 -0.03182906 -0.46597523  0.7479451   0.12580804 -0.36311644
  0.06406616 -0.29641905 -0.39982319 -0.98493049 -0.4431009  -0.29506693
 -0.25931983  0.67831564 -0.80676608 -0.53007712  0.63448069  0.67457029
  0.25457744 -0.82095266 -0.27461275 -0.91244341 -0.69530798 -0.69023852
  0.18840807  0.49891863  0.52417747  0.06833423  0.83449866  0.47608894]

Which suggests to me that the elements of two rows of normvecs you are saying are identical in magnitude are not.

answered Aug 17, 2011 at 12:02

talonmies

72.7k35 gold badges204 silver badges296 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

fedvasu Over a year ago

yes there is no problem , it was an unseen logical error i skipped one line in the code(considering it irrelevant) posted where i had this (wrong, my bad!) assignment tgtvecs[:,0] = np.cos(alphap[:]) ; normvecs[:,] = np.sin(alphap[:])

fedvasu Over a year ago

you would have got the same error , if you had just run my program without modifying it .. anyway thanx

talonmies Over a year ago

No I wouldn't have. It produced no errors as you posted it, all I did after I saw it worked was remove the redundant code that had no effect on the output array you were asking about, and made the output less confusing by printing the difference between the two rows. Otherwise is it precisely as you posted it, sorry.

fedvasu Over a year ago

again you are right!, it would produce no errors , precisely because of the one line removed!!! , thanx again for clearing this up! and yeah i got the whole thing working

Collectives™ on Stack Overflow

difficulty in understanding numpy array behavior

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related