Python/linear_algebra/src/lib.py
Tianyi Zheng a64c9f1e7c
Deduplicate euclidean_length method in Vector (#5658)
* Rewrite parts of Vector and Matrix methods

* Refactor determinant method and add unit tests

Refactor determinant method to create separate minor and cofactor
methods.
Add respective unit tests for new methods.
Rename methods using snake case to follow Python naming conventions.

* Reorganize Vector and Matrix methods

* Update linear_algebra/README.md

Co-authored-by: John Law <johnlaw.po@gmail.com>

* Fix punctuation and wording

* Apply suggestions from code review

Co-authored-by: John Law <johnlaw.po@gmail.com>

* Deduplicate euclidean length method for Vector

* Add more unit tests for Euclidean length method

* Fix bug in unit test for euclidean_length

* Remove old comments for magnitude method

Co-authored-by: John Law <johnlaw.po@gmail.com>
2021-10-31 22:16:02 +08:00

450 lines
14 KiB
Python

"""
Created on Mon Feb 26 14:29:11 2018
@author: Christian Bender
@license: MIT-license
This module contains some useful classes and functions for dealing
with linear algebra in python.
Overview:
- class Vector
- function zero_vector(dimension)
- function unit_basis_vector(dimension, pos)
- function axpy(scalar, vector1, vector2)
- function random_vector(N, a, b)
- class Matrix
- function square_zero_matrix(N)
- function random_matrix(W, H, a, b)
"""
from __future__ import annotations
import math
import random
from typing import Collection, overload
class Vector:
"""
This class represents a vector of arbitrary size.
You need to give the vector components.
Overview of the methods:
__init__(components: Collection[float] | None): init the vector
__len__(): gets the size of the vector (number of components)
__str__(): returns a string representation
__add__(other: Vector): vector addition
__sub__(other: Vector): vector subtraction
__mul__(other: float): scalar multiplication
__mul__(other: Vector): dot product
set(components: Collection[float]): changes the vector components
copy(): copies this vector and returns it
component(i): gets the i-th component (0-indexed)
change_component(pos: int, value: float): changes specified component
euclidean_length(): returns the euclidean length of the vector
angle(other: Vector, deg: bool): returns the angle between two vectors
TODO: compare-operator
"""
def __init__(self, components: Collection[float] | None = None) -> None:
"""
input: components or nothing
simple constructor for init the vector
"""
if components is None:
components = []
self.__components = list(components)
def __len__(self) -> int:
"""
returns the size of the vector
"""
return len(self.__components)
def __str__(self) -> str:
"""
returns a string representation of the vector
"""
return "(" + ",".join(map(str, self.__components)) + ")"
def __add__(self, other: Vector) -> Vector:
"""
input: other vector
assumes: other vector has the same size
returns a new vector that represents the sum.
"""
size = len(self)
if size == len(other):
result = [self.__components[i] + other.component(i) for i in range(size)]
return Vector(result)
else:
raise Exception("must have the same size")
def __sub__(self, other: Vector) -> Vector:
"""
input: other vector
assumes: other vector has the same size
returns a new vector that represents the difference.
"""
size = len(self)
if size == len(other):
result = [self.__components[i] - other.component(i) for i in range(size)]
return Vector(result)
else: # error case
raise Exception("must have the same size")
@overload
def __mul__(self, other: float) -> Vector:
...
@overload
def __mul__(self, other: Vector) -> float:
...
def __mul__(self, other: float | Vector) -> float | Vector:
"""
mul implements the scalar multiplication
and the dot-product
"""
if isinstance(other, float) or isinstance(other, int):
ans = [c * other for c in self.__components]
return Vector(ans)
elif isinstance(other, Vector) and len(self) == len(other):
size = len(self)
prods = [self.__components[i] * other.component(i) for i in range(size)]
return sum(prods)
else: # error case
raise Exception("invalid operand!")
def set(self, components: Collection[float]) -> None:
"""
input: new components
changes the components of the vector.
replaces the components with newer one.
"""
if len(components) > 0:
self.__components = list(components)
else:
raise Exception("please give any vector")
def copy(self) -> Vector:
"""
copies this vector and returns it.
"""
return Vector(self.__components)
def component(self, i: int) -> float:
"""
input: index (0-indexed)
output: the i-th component of the vector.
"""
if type(i) is int and -len(self.__components) <= i < len(self.__components):
return self.__components[i]
else:
raise Exception("index out of range")
def change_component(self, pos: int, value: float) -> None:
"""
input: an index (pos) and a value
changes the specified component (pos) with the
'value'
"""
# precondition
assert -len(self.__components) <= pos < len(self.__components)
self.__components[pos] = value
def euclidean_length(self) -> float:
"""
returns the euclidean length of the vector
>>> Vector([2, 3, 4]).euclidean_length()
5.385164807134504
>>> Vector([1]).euclidean_length()
1.0
>>> Vector([0, -1, -2, -3, 4, 5, 6]).euclidean_length()
9.539392014169456
>>> Vector([]).euclidean_length()
Traceback (most recent call last):
...
Exception: Vector is empty
"""
if len(self.__components) == 0:
raise Exception("Vector is empty")
squares = [c ** 2 for c in self.__components]
return math.sqrt(sum(squares))
def angle(self, other: Vector, deg: bool = False) -> float:
"""
find angle between two Vector (self, Vector)
>>> Vector([3, 4, -1]).angle(Vector([2, -1, 1]))
1.4906464636572374
>>> Vector([3, 4, -1]).angle(Vector([2, -1, 1]), deg = True)
85.40775111366095
>>> Vector([3, 4, -1]).angle(Vector([2, -1]))
Traceback (most recent call last):
...
Exception: invalid operand!
"""
num = self * other
den = self.euclidean_length() * other.euclidean_length()
if deg:
return math.degrees(math.acos(num / den))
else:
return math.acos(num / den)
def zero_vector(dimension: int) -> Vector:
"""
returns a zero-vector of size 'dimension'
"""
# precondition
assert isinstance(dimension, int)
return Vector([0] * dimension)
def unit_basis_vector(dimension: int, pos: int) -> Vector:
"""
returns a unit basis vector with a One
at index 'pos' (indexing at 0)
"""
# precondition
assert isinstance(dimension, int) and (isinstance(pos, int))
ans = [0] * dimension
ans[pos] = 1
return Vector(ans)
def axpy(scalar: float, x: Vector, y: Vector) -> Vector:
"""
input: a 'scalar' and two vectors 'x' and 'y'
output: a vector
computes the axpy operation
"""
# precondition
assert (
isinstance(x, Vector)
and isinstance(y, Vector)
and (isinstance(scalar, int) or isinstance(scalar, float))
)
return x * scalar + y
def random_vector(n: int, a: int, b: int) -> Vector:
"""
input: size (N) of the vector.
random range (a,b)
output: returns a random vector of size N, with
random integer components between 'a' and 'b'.
"""
random.seed(None)
ans = [random.randint(a, b) for _ in range(n)]
return Vector(ans)
class Matrix:
"""
class: Matrix
This class represents an arbitrary matrix.
Overview of the methods:
__init__():
__str__(): returns a string representation
__add__(other: Matrix): matrix addition
__sub__(other: Matrix): matrix subtraction
__mul__(other: float): scalar multiplication
__mul__(other: Vector): vector multiplication
height() : returns height
width() : returns width
component(x: int, y: int): returns specified component
change_component(x: int, y: int, value: float): changes specified component
minor(x: int, y: int): returns minor along (x, y)
cofactor(x: int, y: int): returns cofactor along (x, y)
determinant() : returns determinant
"""
def __init__(self, matrix: list[list[float]], w: int, h: int) -> None:
"""
simple constructor for initializing the matrix with components.
"""
self.__matrix = matrix
self.__width = w
self.__height = h
def __str__(self) -> str:
"""
returns a string representation of this matrix.
"""
ans = ""
for i in range(self.__height):
ans += "|"
for j in range(self.__width):
if j < self.__width - 1:
ans += str(self.__matrix[i][j]) + ","
else:
ans += str(self.__matrix[i][j]) + "|\n"
return ans
def __add__(self, other: Matrix) -> Matrix:
"""
implements matrix addition.
"""
if self.__width == other.width() and self.__height == other.height():
matrix = []
for i in range(self.__height):
row = [
self.__matrix[i][j] + other.component(i, j)
for j in range(self.__width)
]
matrix.append(row)
return Matrix(matrix, self.__width, self.__height)
else:
raise Exception("matrix must have the same dimension!")
def __sub__(self, other: Matrix) -> Matrix:
"""
implements matrix subtraction.
"""
if self.__width == other.width() and self.__height == other.height():
matrix = []
for i in range(self.__height):
row = [
self.__matrix[i][j] - other.component(i, j)
for j in range(self.__width)
]
matrix.append(row)
return Matrix(matrix, self.__width, self.__height)
else:
raise Exception("matrices must have the same dimension!")
@overload
def __mul__(self, other: float) -> Matrix:
...
@overload
def __mul__(self, other: Vector) -> Vector:
...
def __mul__(self, other: float | Vector) -> Vector | Matrix:
"""
implements the matrix-vector multiplication.
implements the matrix-scalar multiplication
"""
if isinstance(other, Vector): # matrix-vector
if len(other) == self.__width:
ans = zero_vector(self.__height)
for i in range(self.__height):
prods = [
self.__matrix[i][j] * other.component(j)
for j in range(self.__width)
]
ans.change_component(i, sum(prods))
return ans
else:
raise Exception(
"vector must have the same size as the "
"number of columns of the matrix!"
)
elif isinstance(other, int) or isinstance(other, float): # matrix-scalar
matrix = [
[self.__matrix[i][j] * other for j in range(self.__width)]
for i in range(self.__height)
]
return Matrix(matrix, self.__width, self.__height)
def height(self) -> int:
"""
getter for the height
"""
return self.__height
def width(self) -> int:
"""
getter for the width
"""
return self.__width
def component(self, x: int, y: int) -> float:
"""
returns the specified (x,y) component
"""
if 0 <= x < self.__height and 0 <= y < self.__width:
return self.__matrix[x][y]
else:
raise Exception("change_component: indices out of bounds")
def change_component(self, x: int, y: int, value: float) -> None:
"""
changes the x-y component of this matrix
"""
if 0 <= x < self.__height and 0 <= y < self.__width:
self.__matrix[x][y] = value
else:
raise Exception("change_component: indices out of bounds")
def minor(self, x: int, y: int) -> float:
"""
returns the minor along (x, y)
"""
if self.__height != self.__width:
raise Exception("Matrix is not square")
minor = self.__matrix[:x] + self.__matrix[x + 1 :]
for i in range(len(minor)):
minor[i] = minor[i][:y] + minor[i][y + 1 :]
return Matrix(minor, self.__width - 1, self.__height - 1).determinant()
def cofactor(self, x: int, y: int) -> float:
"""
returns the cofactor (signed minor) along (x, y)
"""
if self.__height != self.__width:
raise Exception("Matrix is not square")
if 0 <= x < self.__height and 0 <= y < self.__width:
return (-1) ** (x + y) * self.minor(x, y)
else:
raise Exception("Indices out of bounds")
def determinant(self) -> float:
"""
returns the determinant of an nxn matrix using Laplace expansion
"""
if self.__height != self.__width:
raise Exception("Matrix is not square")
if self.__height < 1:
raise Exception("Matrix has no element")
elif self.__height == 1:
return self.__matrix[0][0]
elif self.__height == 2:
return (
self.__matrix[0][0] * self.__matrix[1][1]
- self.__matrix[0][1] * self.__matrix[1][0]
)
else:
cofactor_prods = [
self.__matrix[0][y] * self.cofactor(0, y) for y in range(self.__width)
]
return sum(cofactor_prods)
def square_zero_matrix(n: int) -> Matrix:
"""
returns a square zero-matrix of dimension NxN
"""
ans: list[list[float]] = [[0] * n for _ in range(n)]
return Matrix(ans, n, n)
def random_matrix(width: int, height: int, a: int, b: int) -> Matrix:
"""
returns a random matrix WxH with integer components
between 'a' and 'b'
"""
random.seed(None)
matrix: list[list[float]] = [
[random.randint(a, b) for _ in range(width)] for _ in range(height)
]
return Matrix(matrix, width, height)