replace sets of values in an array without using loops

[paul18fr]

Hi,

I've been thinking I've ever done this, but I do not remember how; in any way I do not understand the Type Error sine I'm using intergers: what I'm missing?
Thanks

import numpy as np

A=np.array([1, 2, 3])

M=np.zeros((10,10), dtype=float)
N=np.copy(M)
P = np.copy(M)

# Explicitly on 1 location
r=3
c=2
M[r, c:c+3]=A

# now on 2 locations
# with a loop
r=np.array([3, 5])
c=np.array([2, 5])
for i in range(2):
    N[r[i], c[i]:c[i]+3]=A

#without a loop
i=np.arange(2)    
j=c[i]+3*np.ones(2, dtype=int)
k = np.ones((2,1), dtype=int)*A
P[r[i], c[i]:j[i]]=k[i]

Error:
TypeError: only integer scalar arrays can be converted to a scalar index

[Larz60+]

you won't get a type error unless their really is one.
Please show complete, unaltered error traceback (always), it includes valuable program execution order information.

[paul18fr]

Hi Larz60+

As requested:

Error:
runfile('C:/forum.py', wdir='C:/MyDir')
Traceback (most recent call last):

  File "C:\outils\Anaconda3\lib\site-packages\spyder_kernels\py3compat.py", line 356, in compat_exec
    exec(code, globals, locals)

  File "c:/forum.py", line 26, in <module>
    P[r[i], c[i]:j[i]]=k[i]

TypeError: only integer scalar arrays can be converted to a scalar index

[Larz60+]

You can write the code as follows to display values if TypeError encountered:

import numpy as np
import sys

A = np.array([1, 2, 3])
M = np.zeros((10, 10), dtype=float)
N = np.copy(M)
P = np.copy(M)

# Explicitly on 1 location
r = 3
c = 2
M[r, c : c + 3] = A
print(f"\nM(r, c:c_3): {M[r, c:c+3]}")

# now on 2 locations
# with a loop
r = np.array([3, 5])
c = np.array([2, 5])

for i in range(2):
    try:
        N[r[i], c[i] : c[i] + 3] = A
    except TypeError:
        print(f"First loop: Values when TypeError occurs:,  i: {i}, c[i]: {c[i]}")
        sys.exit(-1)

# without a loop
i = np.arange(2)
j = c[i] + 3 * np.ones(2, dtype=int)
k = np.ones((2, 1), dtype=int) * A

try:
    P[r[i], c[i] : j[i]] = k[i]
except TypeError:
    print(
        f"\nTypeError, Values when attempting P[r[i], c[i]:j[i]]=k[i]:\n"
        f"    i: {i}, r[i]: {r[i]}, c[i]: {c[i]}, j[i]: {j[i]}"
    )
    sys.exit(-1)

output shows where error occurs:

Output:
M(r, c:c_3): [1. 2. 3.]

TypeError, Values when attempting P[r[i], c[i]:j[i]]=k[i]:
    i: [0 1], r[i]: [3 5], c[i]: [2 5], j[i]: [5 8]

[paul18fr]

Ok I keep it in mind (I know that I need to implement more Exceptions in my code).

Nonetheless I'm here looking for a way to not use a loop to slice following columns; in the example hereafter, it works with a single arugument/tuple, but I cannot go further

Paul

#without a loop
r=np.array([3, 5])
c=np.array([2, 5])
i=np.arange(2)    
d=c[i]+3*np.ones(2, dtype=int)
k = np.ones((2,1), dtype=int)*A
# P[r[i], 2:5]=k[i] # ok
P[r[i], slice(2, 5)]=k[i] # ok

[paul18fr]

After some tests and trials, I can say that the issue comes from the Slice object, but I still trying to understand how to fix it.

import numpy as np
import sys

A=np.array([1, 2, 3])

M=np.zeros((10,10), dtype=float)
N=np.copy(M)
P=np.copy(M)

# # Explicitly on 1 location
r=3
c=2
M[r, c:c+3]=A

# now on 2 locations
# with a loop
r=np.array([3, 5])
c=np.array([2, 5])
for i in range(2):
    N[r[i], c[i]:c[i]+3]=A

#without a loop
i=np.arange(2)    
d=c[i]+3*np.ones(2, dtype=int)
SliceIndexes = np.vstack((c,d))
k = np.ones((2,1), dtype=int)*A
SliceObject = slice(SliceIndexes[i][0], SliceIndexes[i][1])
P[r[i], SliceObject ]=k[i] # ko

# P[r[i], 0:3]=k[i] # ok
# P[r[i], slice(2,5)]=k[i] # ok

If somebody knows how to, it'll be highly appreciated.

Paul

Here bellow the screenshot of information gotten under Spyder:

[Larz60+]

Which version of python are you running?
from command line: python -V

[paul18fr]

Hi Larz60+

Python 3.9.13