NumPy

Python Fundamentals

List/Dictionary/Set Comprehension

<new-list-name> = [ <expression> for <element> in <list-name> if <condition> ]

Class

• private instance variable by adding __ before variable’s name, e.g. self.__x
• class variable changes across object of the same class, e.g. Class.x

NumPy

Integer Array Indexing

a = np.array([[1,2], [3, 4], [5, 6]])
 
# An example of integer array indexing.
# The returned array will have shape (3,) and
print(a[[0, 1, 2], [0, 1, 0]])  # Prints "[1 4 5]"

• Returns an copied 1D array
• Elements can be reused
• Math operators (e.g. +=) for mutating each indexed element

Boolean Array Indexing

Data Types

https://numpy.org/doc/stable/reference/arrays.dtypes.html#arrays-dtypes-constructing

New Axis

arr[np.newaxis] or arr[None]

• Specify new axis position by arr[:, np.newaxis, :], put colon at where axis remains the same
  • Or np.expand_dims(arr, <after_this_dimension>)
• Added to the front, e.g. (5, ) will become (1, 5)

Broadcasting

np.array([1, 2, 3]) is (3, )
np.array([2]) is (1, )

1. Compare shape of array from right to left
2. Prepend 1 to shape if rank is not the same
3. Broadcasting is doable if all dimensions is compatible (same or either one is 1)

Important

p. 80-82, 87-88

https://www.pythonlikeyoumeanit.com/Module3_IntroducingNumpy/Broadcasting.html

Centering

1. Calculate mean of columns arr.mean(<dimension>)
2. Subtract mean from each entry to obtain the centered array

Pairwise Distances

https://stackoverflow.com/a/37903795

Cheatsheets

np.array(<list>)
np.zeros(<shape>)
np.ones(<shape>)
 
np.arange(<range>)
np.linspace(start=, stop=, num=)
np.logspace(start=, stop=, num=)
 
<arr2> = <arr1>.copy()

<arr>.shape
<arr>.ndim
<arr>.size