Use Python Numeric Variables in MATLAB

Use Python Numeric Types in MATLAB

By default, a number in MATLAB has a double-precision type. However, in Python, a number without a fractional part has, by default, an integer type. To resolve conversion errors when calling a Python function that takes a numeric input argument, MATLAB converts double values into types that best represent the data to the Python language.

For example, to call trigonometry functions in the Python math module, pass a MATLAB double value.

pynum = py.math.radians(90)

pynum = 
1.5708

For functions that return a Python float type, MATLAB automatically converts the type to double.

class(pynum)

ans = 
'double'

For functions that return a Python int type, MATLAB automatically converts the type to int. For example, the bit_length function returns the number of bits necessary to represent an integer in binary as an int value.

py.int(intmax).bit_length

ans = 
  Python int with properties:

    denominator: [1×1 py.int]
           imag: [1×1 py.int]
      numerator: [1×1 py.int]
           real: [1×1 py.int]

    31

MATLAB also uses Python type hints when converting MATLAB double values to Python int before passing arguments to a Python function. (since R2024b)

For example, the Python date constructor requires integer inputs. When you pass MATLAB numbers to the Python date constructor in the datetime module, MATLAB first converts the double values to integers.

d = py.datetime.date(2014,12,31)

d = 
  Python date with properties:

      day: [1×1 py.int]
    month: [1×1 py.int]
     year: [1×1 py.int]

    2014-12-31

The conversion from double to int is also supported for user-authored functions that provide type hint information. However, the use of type hint information in MATLAB is limited. For example, modules such as the Python typing module are not supported. MATLAB also does not support type hints from stub modules or modules containing stubs. Further, MATLAB converts double to int only if there is no precision loss.

Call Python Methods with Numeric iterable Arguments

The Python math.fsum function sums floating-point values in an iterable input argument. You can pass a MATLAB vector to this function. For example, open the MATLAB patients.mat data file and read the numeric array Height.

load patients.mat
class(Height)

ans = 
'double'

size(Height)

ans = 1×2

   100     1

When you pass this argument to Python, MATLAB automatically converts the numeric values to Python numeric values and Python iterates over the vector values.

py.math.fsum(Height)

ans = 
6707

Use Python array Types in MATLAB

Suppose that you have a Python function that returns the following Python array.array of type double.

P = py.array.array('d', 1:5)

P = 
  Python array:

     1     2     3     4     5

    Use details function to view the properties of the Python object.

    Use double function to convert to a MATLAB array.

To pass P to the MATLAB function sum, convert P to a MATLAB array of type double.

sum(double(P))

ans = 
15

Use Python Integer array Types in MATLAB

Suppose that you have this Python array. Call the Python reverse function on the array, and then convert the result to a MATLAB array.

arr = py.array.array('i',[int32(5),int32(1),int32(-5)])

arr = 
  Python array:

    5    1   -5

    Use details function to view the properties of the Python object.

    Use int32 function to convert to a MATLAB array.

arr.reverse
A = int32(arr)

A = 1×3 int32 row vector

   -5    1    5

Why Do I See Properties When I Display a Number?

MATLAB displays all Python types as objects, which includes a list of object properties. For numeric types, MATLAB displays the expected output value on the last line.

py.int(5)

ans = 
  Python int with properties:

    denominator: [1×1 py.int]
           imag: [1×1 py.int]
      numerator: [1×1 py.int]
           real: [1×1 py.int]

    5