PyTorch Basics | Part One | Scalars, Arrays, and Matrix

PyTorch is a popular open-source machine learning library developed by the Facebook AI Research lab. It is widely used for various applications such as computer vision, natural language processing, and many other machine learning tasks. In this article, we will cover the basics of PyTorch, focusing on Scalars, Arrays, and Matrix.

Scalars

A scalar in PyTorch is a single value, which can be a number or a variable representing a number. Scalars are created using the torch.tensor class, which is the fundamental data structure in PyTorch. For example:


import torch

# Creating a scalar
scalar = torch.tensor(5)


Scalars can be used for simple arithmetic operations, such as addition, subtraction, multiplication, and division. PyTorch provides efficient functions for these operations, making it easy to perform calculations with scalars.

Arrays

Arrays in PyTorch are multi-dimensional collections of numbers. They can be created using the torch.tensor class with a list of numbers or nested lists. For example:


# Creating an array
array = torch.tensor([1, 2, 3, 4])


Arrays can be used to represent vectors, which are one-dimensional arrays, or matrices, which are two-dimensional arrays. PyTorch provides a wide range of functions for manipulating arrays, such as reshaping, transposing, and concatenating.

Matrix

A matrix in PyTorch is a two-dimensional array with rows and columns. It can be created using the torch.tensor class with a nested list of numbers. For example:


# Creating a matrix
matrix = torch.tensor([[1, 2], [3, 4], [5, 6]])


Matrices are commonly used in machine learning for representing data, such as images, text, and numerical features. PyTorch provides powerful matrix operations, such as matrix multiplication, element-wise multiplication, and matrix inversion.

These are the basic data structures in PyTorch, which form the foundation for building machine learning models. In the next part of this series, we will delve into tensors, operations, and gradients in PyTorch.