Matlab struct array to python

Question

I have thoroughly looked around to try figuring out a way to create a matlab like struct array in python. There are some questions online that I looked at and either the answers don't seem to help or I might simply be misinterpreting them as they pertain to me. So, moving on. I am trying to form a python equivalent to the following matlab code.

channel                 = [];
channel.PRN             = 0; 
channel.acquiredFreq    = 0; 
channel.codePhase       = 0; 
channel.status          = '-';  
channel = repmat(channel, 1, settings.numberOfChannels);

Where repmat would basically create a struct array called channel with a number of cells equal to settings.numberOfChannels and each of those would have PRN,acquiredFreq, etc.

Later on, I access this struct by performing a loop that alters these values as such:

for ii = 1:settings.numberOfChannels
        channel(ii).PRN          = PRNindexes(ii);
        channel(ii).acquiredFreq = acqResults.carrFreq(PRNindexes(ii));
        channel(ii).codePhase    = acqResults.codePhase(PRNindexes(ii));

I have tried several approaches but it either spits out nonsense in the case of tile using numpy(which i might have just been using incorrectly) or when I attempt to make a loop such as:

class test:
    for iii in range(1,settings.numberOfChannels):
        iii.PRN=0
        iii.acquiredFreq=0
        iii.codePhase=0
        iii.status="-"

More than likely I suppose it's a syntax error or my misunderstanding of python since this is my first time utilizing it. If this is the incorrect place to ask this or something of that nature, I apologize.

Thank you

unutbu · Accepted Answer · 2018-02-12 14:40:36Z

Update: You may want to investigate Pandas. Its Series and DataFrames are easier to work with and more full-featured than NumPy structured arrays.

You could use a NumPy structured array:

import numpy as np
channel = np.zeros(1, dtype = [('PRN',int),
                               ('acquiredFreq',int),
                               ('codePhase',int),
                               ('status','|S1')])

print(channel)
# [(0, 0, 0, '')]

Indexing by integer accesses a particular row:

print(channel[0])
# (0, 0, 0, '')

Indexing by column name returns the column as an array:

print(channel['PRN'])
# [0]

Or you can loop through each row and assign to each field (column), but this is relatively slow in NumPy.

for row in channel:
    row['PRN'] = 1
    row['acquiredFreq'] = 1
    row['codePhase'] = 1
    row['status'] = '+'

print(channel)    
# [(1, 1, 1, '+')]

Just for completeness, I'll also mention you can assign by row then column:

channel[0]['status'] = '-'
print(channel)
# [(1, 1, 1, '-')]

or assign by column then row:

channel['PRN'][0] = 10
print(channel)
# [(10, 1, 1, '-')]

I showed the above because it is most similar to the Matlab code you posted. However, let me stress again that assigning to individual cells in a NumPy array is slow. The NumPy-way to do the above is to do whole-array assignments instead:

channel['PRN'] = PRNindexes

where PRNindexes is a sequence (e.g. a list, tuple, or NumPy array).

You can also use fancy indexing (aka "advanced indexing") to select rows:

index = (channel.status == '+')  # Select all rows with status '+'
channel['PRN'][index] = 10       # Set PRN to 10 for all those rows

Just keep in mind that fancy indexing returns a new array, not a view of the original array. (In contrast, "basic slicing" (e.g. channel[0] or channel[1:10] returns a view.) So if you want to make assignments that alter the original array, select by column first, then fancy index (index)

channel['PRN'][index] = ...

rather than

channel[index]['PRN'] = ...

This was exactly what I needed..after spending so much time you managed to alleviate all my frustrations. Thank you!

Collectives™ on Stack Overflow

Matlab struct array to python

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related