Convert array 1:n in matlab to python

Question

I have question similar to this, and possibly a much simpler one:

We use the

k = 1:n;

a lot in Matlab. But in Python I simply struggle to get this simple thing work. I tried using arange, but never really realized what the error is:

>>> arange(1,]10[,1,])
File "<stdin>", line 1
arange(1,]10[,1,])
         ^
SyntaxError: invalid syntax
>>> arange(1,10[,1,])
  File "<stdin>", line 1
    arange(1,10[,1,])
                ^
SyntaxError: invalid syntax
>>> arange([1,]10[,1,])
  File "<stdin>", line 1
    arange([1,]10[,1,])
                ^
SyntaxError: invalid syntax
>>> np.arange
<built-in function arange>
>>> arange([1], 10[,1])
  File "<stdin>", line 1
    arange([1], 10[,1])
                   ^
SyntaxError: invalid syntax

I went to the numpy website and tried to give the syntax there, but again:

>>> import numpy as np
>>> np.arange([1],10[,1],dtype=None)
  File "<stdin>", line 1
    np.arange([1],10[,1],dtype=None)
                     ^
SyntaxError: invalid syntax

I wouldn't have written this post just to clarify something of this sort, but my point is why is this simplest Matlab command so very complicated in Python? I even used this tool to convert .m codes to .py codes, with little effect.

EDIT after the post from @mskimm: Thanks! One related question. To write something very similar to the following in Matlab:

n = 100;
k = 1:n;
z = (n-k)./(n-k-1);
plot(k,log(z))

I ended up writing this in Python:

from __future__ import division
import numpy as np
import matplotlib.pyplot as plt

# test code to check the shape of a curve
n = 100
k = np.arange(1,n+1)
z = (n - k) / (n - k - 1)

fig = plt.figure()
ax = fig.gca()

plt.plot(k, np.log(z))
plt.show()

Is this the right way of doing it? Or there is a better way of doing the plot() ?

Python is not more complicated than matlab. It has the added advantage that one can know what it is doing/intends to do at any given moment. The main big difference to watch out here for is that python arrays naturally start at 0 and go to n-1 (standard for most programming languages), whereas matlab (and also R) start indexing at 1 and end at n. Take great care of this difference. As for the solution, the previous comment indicates it well. Take range(n) if you want it to be more simple. Take range(1, n+1) if you really need it to go from 1 to n — eickenberg
– eickenberg, Commented Apr 15, 2014 at 9:47
the code includes divide by 0 and log(inf). Do you intend to do? — emesday
– emesday, Commented Apr 15, 2014 at 12:00
If replacing k = np.arange(1,n+1) with k = np.arange(1,n-1). It works. — emesday
– emesday, Commented Apr 15, 2014 at 12:27
Yes, I could have avoided the division myself, but was a little lazy as Matlab usually takes care of this thing and does not show those points if encountered. But the main question is: is this the right (and minimal) piece of code? I now realize that I could omit the fig = plt.figure() and ax = fig.gca() but want to know if the rest 5 lines is the minimal replacement for the 4 line Matlab code. — Pagol
– Pagol, Commented Apr 17, 2014 at 2:09

Glorfindel · Accepted Answer · 2023-01-09 20:03:47Z

1

Welcome to Numpy!

1:n is the same as arange(1, n+1, 1) where the first 1 is start, n+1 is stop, and the last 1 is step. Note that Numpy index is started at 0 whereas Matlab is 1. So I think

np.arange(0, n, 1) # will generate 0, 1, ..., n-1

is more appropriate to array indexing.

[] means optional value. If you don't give an explicit value, start is 0 and step is 1. You can use

np.arange(n) # will generate 0, 1, ..., n-1

This page helps Matlab users.

UPDATE

Comment for Matlab Users who start learning Numpy and is familiar with Matlab IDE.

I recommend to use Spyder or IPython Notebook.

Spyder IPython Notebook

edited Jan 9, 2023 at 20:03

Glorfindel

22.8k13 gold badges97 silver badges124 bronze badges

answered Apr 15, 2014 at 9:51

emesday

6,2063 gold badges31 silver badges46 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Pagol Over a year ago

Thanks and perhaps you can clarify my doubt about the plot (after the EDIT in the main question)

TheBlackCat · Accepted Answer · 2014-05-05 13:26:10Z

For your first question, the simplest ways to do this in numpy are:

import numpy as np
n = 5
k = np.r_[1:n+1]

and:

import numpy as np
n = 5
k = np.r_[:n]+1

The first approach is a tiny bit faster.

For your second question, I see you already found out that you can leave out most of it. There is one additional thing you can do. By default, matplotlib does not update the plot until you do "plt.show()". However, you can change this by doing something like this:

from __future__ import division
import numpy as np
import matplotlib.pyplot as plt

plt.interactive(True)

# test code to check the shape of a curve
n = 100
k = np.r_[1:n+1]
z = (n - k) / (n - k - 1)

plt.plot(k, np.log(z))

You only have to do this once when you first start, and you can set it to False at and point later.

You can also configure matplotlib to do this by default by opening your matplotlibrc file and changing:

#interactive  : False

to

interactive  : True

However, you are better of using IPython, which handles all of this for if you run it using "ipython --pylab":

from __future__ import division

# test code to check the shape of a curve
n = 100
k = r_[1:n+1]
z = (n - k) / (n - k - 1)

plot(k, log(z))

Collectives™ on Stack Overflow

Convert array 1:n in matlab to python

2 Answers 2

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related