matlab to python code conversion

Question

I am trying to convert a matlab code to python due to lack of matlab. I will be grateful if you kindly tell me the python equivalents of the following functions which I could not find:

letter2number_map('A') = 1;
number2letter_map(1) = 'A';
str2num()
strcmp()
trace()
eye()
getenv('STRING')
[MI_true, ~, ~] = function() What does ~ mean?
mslice
ones()

Thank you very much for your kind assistance.

• Tal Darom: If you lack matlab why not give Octave or SciLab a chance?
• tacaswell: see scipy.org/NumPy_for_Matlab_Users
• 0 _: Related: stackoverflow.com/questions/9845292/…

Answer 1

I don't actually know matlab, but I can answer some of those (assuming you have numpy imported as np):

letter2number_map('A') = 1; -> equivalent to a dictionary: {'A':1} 
number2letter_map(1) = 'A'; -> equivalent to a dictionary: {1:'A'}
str2num() -> maybe float(), although a look at the docs makes 
 it look more like eval -- Yuck. 
 (ast.literal_eval might be closest)
strcmp() -> I assume a simple string comparison works here. 
 e.g. strcmp(a,b) -> a == b
trace() -> np.trace() #sum of the diagonal
eye() -> np.eye() #identity matrix (all zeros, but 1's on diagonal)
getenv('STRING') -> os.environ['STRING'] (with os imported of course)
[MI_true, ~, ~] = function() -> Probably: MI_true,_,_ = function() 
 although the underscores could be any 
 variable name you wanted. 
 (Thanks Amro for providing this one)
mslice -> ??? (can't even find documentation for that one)
ones() -> np.ones() #matrix/array of all 1's

• Amro: mslice is not a built-in MATLAB function. ~ is for ignoring function return arguments
• mgilson: @Amro -- did I get the rest of the table approximately correct? (thanks by the way)

Answer 2

Converting from letter to number: ord("a") = 97

Converting from number to letter: chr(97) = 'a'

(You can subtract 96 to get the result you're looking for for lowercase, or 64 for uppercase.)

Parse a string to int: int("523") = 523

Comparing strings (Case sensitive): "Hello"=="Hello" = True

Case insensitive: "Hello".lower() == "hElLo".lower() = True

ones(): [1]*ARRAY_SIZE

Identity matrix: [[int(x==y) for x in range(5)] for y in range(5)]

To make 2-D arrays, you need to use numpy.

Edit:

Alternatively, you can make a 5x5 2-D array like so: [[1 for x in range(5)] for y in range(5)]

• mgilson: In this, trace could be implemented sum(a[i][i] for i in range(len(a)))
• tacaswell: If the op is trying to replace matlab, I think assuming numpy is fair.