CODEWITHSUNDEEP
pythonnumpymatplotlib
Mikael Fremling
Mikael Fremling

Reputation: 762

matplotlib: "TypeError: Image data can not convert to float" for what look like a fine matrix

I know that other people are seeing similar errors (TypeError: Image data can not convert to float, TypeError: Image data can not convert to float using matplotlib, Type Error: Image data can not convert to float) but i don't see any solution there that helps me.

I'm trying to populate a numpy-array with floating point data and the plot it using imshow. The data in the Y-direction (almost) a Hermite polynomial and a Gaussian envelope, whereas the X-direction is just a Gaussian envelope.

from __future__ import print_function
import numpy as np
import matplotlib.pyplot as plt

####First we set Ne
Ne=25

###Set up a mesh with size sqrt(Ne) X sqrt(Ne)
sqrtNe=int(np.sqrt(Ne))
Ky=np.array(range(-sqrtNe,sqrtNe+1),dtype=float)
Kx=np.array(range(-sqrtNe,sqrtNe+1),dtype=float)
[KXmesh,KYmesh]=np.meshgrid(Kx,Ky,indexing='ij')

##X-direction is gussian envelope
AxMesh=np.exp(-(np.pi*KXmesh**2)/(4.0*Ne))

Nerror=21 ###This is where the error shows up
for n in range(Nerror,Ne):
    ##Y-direction is a polynomial of degree n ....
    AyMesh=0.0
    for i in range(n/2+1):
        AyMesh+=(-1)**i*(np.sqrt(2*np.pi)*2*KYmesh)**(n-2*i)/(np.math.factorial(n-2*i)*np.math.factorial(i))
    ### .... times a gaussian envelope
    AyMesh=AyMesh*np.exp(-np.pi*KYmesh**2)
    AyMesh=AyMesh/np.max(np.abs(AyMesh))
    WeightMesh=AyMesh*AxMesh
    print("n:",n)
    plt.figure()
    ####Error occurs here #####
    plt.imshow(WeightMesh,interpolation='nearest')
    plt.show(block=False)

When the code reaches the impow then i get the following error message

Traceback (most recent call last):
  File "FDOccupation_mimimal.py", line 30, in <module>
    plt.imshow(WeightMesh,interpolation='nearest')
  File "/usr/lib/python2.7/dist-packages/matplotlib/pyplot.py", line 3022, in imshow
    **kwargs)
  File "/usr/lib/python2.7/dist-packages/matplotlib/__init__.py", line 1814, in inner
    return func(ax, *args, **kwargs)
  File "/usr/lib/python2.7/dist-packages/matplotlib/axes/_axes.py", line 4947, in imshow
    im.set_data(X)
  File "/usr/lib/python2.7/dist-packages/matplotlib/image.py", line 449, in set_data
    raise TypeError("Image data can not convert to float")
TypeError: Image data can not convert to float

If i replace the code 

AyMesh=0.0
for i in range(n/2+1):
    AyMesh+=(-1)**i*(np.sqrt(2*np.pi)*2*KYmesh)**(n-2*i)/(np.math.factorial(n-2*i)*np.math.factorial(i))
### .... times a gaussian envelope
AyMesh=AyMesh*np.exp(-np.pi*KYmesh**2)
AyMesh=AyMesh/np.max(np.abs(AyMesh))

with simply 

AyMesh=KYmesh**n*np.exp(-np.pi*KYmesh**2)
AyMesh=AyMesh/np.max(np.abs(AyMesh))

the problem goes away!? Does anyone understand what is happening here?

Upvotes: 2

Views: 27086

Answers (1)

Pierre de Buyl
Pierre de Buyl

Reputation: 7293

For large values, np.math.factorial returns a long instead of an int. Arrays with long values are of dtype object as the cannot be stored using NumPy's types. You can re-convert the final result by 

WeightMesh=np.array(AyMesh*AxMesh, dtype=float)

to have a proper float array.

Upvotes: 7

Related Questions