Problem interpolating using RBF from python's scipy

Question

I am following the accepted answer of this thread using my ownd gridded data.

I load it as:

import numpy as np
import matplotlib.pyplot as plt
from scipy import interpolate
import pylab as py

token = open('Ydata_48_of_50.txt','r')
linestoken=token.readlines()
tokens_column_numberX = 0

resulttokenX=[]

for x in linestoken:
    resulttokenX.append(x.split()[tokens_column_numberX])
token.close()

resulttokenX = np.array(resulttokenX)

(I do the same for Y and F(X, Y)) and then, I use what is displayed in the aforementioned link:

xi, yi = np.linspace(resulttokenX.min(), resulttokenX.max(), 200), np.linspace(resulttokenY.min(), resulttokenY.max(), 200)
xi, yi = np.meshgrid(xi, yi)

# Interpolate
rbf = scipy.interpolate.Rbf(resulttokenX, resulttokenY, resulttokenF, function='linear')

Unfortunately, the last line here is an error. I get

    xi, yi = np.linspace(resulttokenX2.min(), resulttokenX2.max(), 200), np.linspace(resulttokenY2.min(), resulttokenY2.max(), 200)

  File "D:\Users\me\anaconda3\lib\site-packages\numpy\core\_methods.py", line 43, in _amin
    return umr_minimum(a, axis, None, out, keepdims, initial, where)

TypeError: cannot perform reduce with flexible type

I have no idea why this happens, since in the original code x appears in the last line and is

type(x)
Out[26]: numpy.ndarray

which is the same type of variable as

type(resulttokenX2)
Out[24]: numpy.ndarray

I don't know why this happens. Can someone tell me what I have to do to reproduce the original code with my gridded data instead of random?

Thanks.

Edit:

resulttokenY2
Out[3]: 
array(['3.2000000e+01', '3.2000000e+01',

are the first lines of resulttokenY2

@yann ziselman x.dtype is float64 and resulttokenX2.dtype is ' — slow_learner, Oct 11 '21 at 10:28
try this: xi, yi = np.linspace(resulttokenX **.astype(float)** .min(), resulttokenX **.astype(float)** .max(), 200), np.linspace(resulttokenY **.astype(float)** .min(), resulttokenY **.astype(float)** .max(), 200) — yann ziselman, Oct 11 '21 at 10:30
would you kindly show us a **small** portion of the content of resulttokenY? — yann ziselman, Oct 11 '21 at 10:31
I have tested what you suggested and it works, albeit I needed some extra work in the plotting part. I will edit my question to display some elements of resulttokenY — slow_learner, Oct 11 '21 at 10:38
None. It works fine. Thanks. I edited since you asked, but everything works now — slow_learner, Oct 11 '21 at 10:40
check [this](https://stackoverflow.com/help/self-answer) out — yann ziselman, Oct 11 '21 at 10:41
yep, you can clearly see in the portion of resulttokenY2 that the dtype was string and that caused the TypeError. next time try looking at your data. that might solve your issue faster than asking a quesiton on SO and save you some time — yann ziselman, Oct 11 '21 at 10:43

score 0 · Accepted Answer · answered Oct 11 '21 at 10:44

After some very helpful advice from Yann ziselman I have managed to do it. This is the full code:

import numpy as np
import matplotlib.pyplot as plt
from scipy import interpolate
import pylab as py
import scipy

token = open('Ydata_48_of_50.txt','r')
linestoken=token.readlines()
tokens_column_numberX = 0
tokens_column_numberY = 1
tokens_column_numberF = 2

resulttokenX=[]
resulttokenY=[]
resulttokenF=[]
for x in linestoken:
    resulttokenX.append(x.split()[tokens_column_numberX])
    resulttokenY.append(x.split()[tokens_column_numberY])
    resulttokenF.append(x.split()[tokens_column_numberF])
token.close()


resulttokenX2 = np.array(resulttokenX)
resulttokenY2 = np.array(resulttokenY)
resulttokenF2 = np.array(resulttokenF)



# Set up a regular grid of interpolation points
xi, yi = np.linspace(resulttokenX2.astype('float').min(), resulttokenX2.astype('float').max(), 100), np.linspace(resulttokenY2.astype('float').min(), resulttokenY2.astype('float').max(), 100)
xi, yi = np.meshgrid(xi, yi)

# Interpolate
rbf = scipy.interpolate.Rbf(resulttokenX2, resulttokenY2, resulttokenF2, function='linear')
zi = rbf(xi, yi)

plt.imshow(zi, vmin=resulttokenF2.astype('float').min(), vmax=resulttokenF2.astype('float').max(), origin='lower', extent=[resulttokenX2.astype('float').min(), resulttokenX2.astype('float').max(), resulttokenY2.astype('float').min(), resulttokenY2.astype('float').max()])
plt.scatter(resulttokenX2.astype('float'), resulttokenY2.astype('float'), c=resulttokenF2.astype('float'))
plt.colorbar()
plt.show()

Problem interpolating using RBF from python's scipy

1 Answers1