python数组对列的行值进行操作

Question

I have a array

array([[  1.00000000e+00,   1.08020000e+04,   1.08070000e+04],
       [  1.00000000e+00,   1.08760000e+04,   1.08810000e+04],
       [  1.00000000e+00,   1.09520000e+04,   1.09570000e+04],
       [  1.00000000e+00,   1.18130000e+04,   1.18180000e+04],
       [  1.00000000e+00,   1.21400000e+04,   1.21450000e+04],
       [  1.00000000e+00,   1.26300000e+04,   1.26350000e+04],
       [  1.00000000e+00,   1.50800000e+04,   1.50850000e+04],
       [  1.00000000e+00,   1.56590000e+04,   1.56640000e+04],
       [  1.00000000e+00,   1.70740000e+04,   1.70790000e+04]])

I have to subtract row1 and row2 of column 2 and add the result to a new column in row1. May someone can help achieving this.

result1 =row2-row1
result2 =row3-row2

I want to iterate in a loop for all values in column 2.

I tried to write this but not able to perform correct slicing:

testfile = numpy.loadtxt("numpy_test.txt", skiprows=1, usecols=(0,1,2))
i=1
j=i-1
k=1 
for array in testfile:
    print array[i][k]-array[j][k] #result = arr[1][1] - arr[0][1] 
    i+=1
    j+=1 

How to iterate over a loop that is my question, already you helped me in understanding how we may code for elements.

Answer 1

i think my answer is not good enough. However, i cannot comment, so i have to put my thought in the answer.

Let your array be a. My solution is:

for item in a:
    item[0] = item[1] - item[0] 
    # you can add item[1] = item[2] - item[1] if you want

and this will return a with the result.

Answer 2

I think you want something like this?:

array = [[  1.00000000e+00,   1.08020000e+04,   1.08070000e+04],
   [  1.00000000e+00,   1.08760000e+04,   1.08810000e+04],
   [  1.00000000e+00,   1.09520000e+04,   1.09570000e+04],
   [  1.00000000e+00,   1.18130000e+04,   1.18180000e+04],
   [  1.00000000e+00,   1.21400000e+04,   1.21450000e+04],
   [  1.00000000e+00,   1.26300000e+04,   1.26350000e+04],
   [  1.00000000e+00,   1.50800000e+04,   1.50850000e+04],
   [  1.00000000e+00,   1.56590000e+04,   1.56640000e+04],
   [  1.00000000e+00,   1.70740000e+04,   1.70790000e+04]]

result1 = array[1][1] - array[0][1] # 1.08760000e+04 - 1.08020000e+04
result2 = array[2][1] - array[1][1] # 1.09520000e+04 - 1.08760000e+04

Answer 3

Numpy matrices can be accessed like multi dimensional arrays:

result1 = arr[1][1] - arr[0][1]
result2 = arr[2][1] - arr[1][1]

To add a new column create a new temp array, and after you have copied arr 's contents to it, copy all the contents back to arr :

temp = np.zeros((arr.shape[0], arr.shape[1]+1))
temp[:,:-1] = arr
temp[0][temp.shape[1]+1] = result1 + result2

All in all:

import numpy as np
arr = np.array([[  1.00000000e+00,   1.08020000e+04,   1.08070000e+04],
       [  1.00000000e+00,   1.08760000e+04,   1.08810000e+04],
       [  1.00000000e+00,   1.09520000e+04,   1.09570000e+04],
       [  1.00000000e+00,   1.18130000e+04,   1.18180000e+04],
       [  1.00000000e+00,   1.21400000e+04,   1.21450000e+04],
       [  1.00000000e+00,   1.26300000e+04,   1.26350000e+04],
       [  1.00000000e+00,   1.50800000e+04,   1.50850000e+04],
       [  1.00000000e+00,   1.56590000e+04,   1.56640000e+04],
       [  1.00000000e+00,   1.70740000e+04,   1.70790000e+04]])
result1 = arr[1][1] - arr[0][1]
result2 = arr[2][1] - arr[1][1]
temp = np.zeros((arr.shape[0], arr.shape[1]+1))
temp[:,:-1] = arr
temp[0][temp.shape[1]-1] = result1 + result2
print temp

Gives:

[[  1.00000000e+00   1.08020000e+04   1.08070000e+04   1.50000000e+02]
 [  1.00000000e+00   1.08760000e+04   1.08810000e+04   0.00000000e+00]
 [  1.00000000e+00   1.09520000e+04   1.09570000e+04   0.00000000e+00]
 [  1.00000000e+00   1.18130000e+04   1.18180000e+04   0.00000000e+00]
 [  1.00000000e+00   1.21400000e+04   1.21450000e+04   0.00000000e+00]
 [  1.00000000e+00   1.26300000e+04   1.26350000e+04   0.00000000e+00]
 [  1.00000000e+00   1.50800000e+04   1.50850000e+04   0.00000000e+00]
 [  1.00000000e+00   1.56590000e+04   1.56640000e+04   0.00000000e+00]
 [  1.00000000e+00   1.70740000e+04   1.70790000e+04   0.00000000e+00]]

Answer 4

This is how I would do it:

import numpy as np

A = np.array([[ 1.0,  1.0802e4,  1.0807e4],
              [ 1.0,  1.0876e4,  1.0881e4],
              [ 1.0,  1.0952e4,  1.0957e4],
              [ 1.0,  1.1813e4,  1.1818e4],
              [ 1.0,  1.2140e4,  1.2145e4],
              [ 1.0,  1.2630e4,  1.2635e4],
              [ 1.0,  1.5080e4,  1.5085e4],
              [ 1.0,  1.5659e4,  1.5664e4],
              [ 1.0,  1.7074e4,  1.7079e4]])

# First add a column to the array
m, n = A.shape
new_column = np.zeros((m, 1))
A = np.hstack((A, new_column))

# Now loop over the rows and calculate differences
for i in range(m-1):
    A[i, 3] = A[i+1, 1] - A[i, 1]

# Or alternatively, without a loop
A[:-1, 3] = np.diff(A[:, 1])

Notice that you can't iterate the Numpy array directly as for array in testfile because you need values from multiple rows.