CODEWITHSUNDEEP
pythonarraysnumpy
okvoyce
okvoyce

Reputation: 153

Find indices of element in 2D array

I have a piece of code below that calculates the maximum value of an array. It then calculates a value for 90% of the maximum, finds the closest value to this in the array as well as its corresponding index.

I need to ensure that I am finding the closest value to 90% that occurs only before the maximum. Can anyone help with this please? I was thinking about maybe compressing the array after the maximum has occurred but then each array I use will be a different size and that will be difficult later on. 

import numpy as np

#make amplitude arrays
amplitude=[0,1,2,3, 5.5, 6,5,2,2, 4, 2,3,1,6.5,5,7,1,2,2,3,8,4,9,2,3,4,8,4,9,3]

#split arrays up into a line for each sample
traceno=5                  #number of traces in file
samplesno=6                #number of samples in each trace. This wont change.

amplitude_split=np.array(amplitude, dtype=np.int).reshape((traceno,samplesno))

#find max value of trace
max_amp=np.amax(amplitude_split,1)

#find index of max value
ind_max_amp=np.argmax(amplitude_split, axis=1, out=None)

#find 90% of max value of trace
amp_90=np.amax(amplitude_split,1)*0.9

# find the indices of the min absolute difference 
indices_90 = np.argmin(np.abs(amplitude_split - amp_90[:, None]), axis=1)
print("indices for 90 percent are", + indices_90)

Upvotes: 0

Views: 111

Answers (1)

Tls Chris
Tls Chris

Reputation: 3934

Use a mask to set the values after the maximum (including the maximum? ) to a known 'too high' value. Then argmin will return the index of the minimum difference in the 'valid' area of each row.

# Create a mask for amplitude equal to the maximum  
# add a dimension to max_amp.  
mask = np.equal(amplitude_split, max_amp[-1, None]) 

# Cumsum the mask to set all elements in a row after the first True to True 
mask[:] = mask.cumsum(axis = 1)
mask
# array([[False, False, False, False, False,  True],
#        [ True,  True,  True,  True,  True,  True],
#        [False, False, False,  True,  True,  True],
#        [False, False, False, False,  True,  True],
#        [False, False, False, False,  True,  True]])

# Set inter to the absolute difference.
inter = np.abs(amplitude_split - amp_90[-1,None])

# Set the max and after to a high value (10. here).
inter[mask] = max_amp.max()   # Any suitably high value

inter  # Where the mask is True inter == 9. 
# array([[8.1, 7.1, 6.1, 5.1, 3.1, 9. ],
#        [9. , 9. , 9. , 9. , 9. , 9. ],
#        [7.1, 2.1, 3.1, 9. , 9. , 9. ],
#        [6.1, 5.1, 0.1, 4.1, 9. , 9. ],
#        [5.1, 4.1, 0.1, 4.1, 9. , 9. ]])

# Find the indices of the minimum in each row
np.argmin(inter, axis = 1)
# array([4, 0, 1, 2, 2])

Upvotes: 2

Related Questions