A Tensor Operations Library in C

This library provides an implementation for creating, manipulating, and performing arithmetic operations on tensors in C. This library uses my own 'c utlities library' as a dependency, except for which everything is self-dependent. Below is an overview of the key features and functionality.

Uses

Features

1. Tensor Creation

• Create tensors with specified dimensions and initialize them with same value in heap.
• Example:

Tensor t = tensor_create(allocr, 1.f, 2,3,4);

• Supports indexing of tensor elements directly via macros as:

tensor_inx(t, 1,1,1) = 34;
tensor_inx(t, 0,0,0) = 12;

2. Tensor Printing

• Print tensors in a human-readable format for debugging or visualization.
• Example:

tensor_print(allocr, t);

3. Tensor Permutation

• Change the order of dimensions in a tensor using permutation.
• Example:

tensor_permute(t, 1, 2);

4. Arithmetic Operations on Tensors

Perform element-wise operations on tensors:

• Addition: Add two tensors element-wise.

Tensor t3 = tensor_add(allocr, t1, t2);

• Multiplication: Multiply two tensors element-wise.

Tensor t5 = tensor_prod(allocr, t3, t4);

• Generate random tensors within specified ranges:

Tensor t1 = tensor_random(allocr, 0.f, 100.f, DIM);

5. Creating Tensors on Stack

• Create and initialize stack-based tensors with multi-dimensional data.
• Example:

Tensor t = MAKE_STACK_TENSOR(({{{1,2}, {2,3}, {4,5}},
{{-3,2}, {-8,7}, {0,-9}},
{{1,8}, {4,-2}, {7,-3}}}),
3,4,2);

6. Contiguous Tensor Conversion

• Convert tensors into contiguous memory layout for efficient storage and access.
• Example:

Tensor dt = tensor_contiguous(allocr, t);

7. Tensor Storage Information

• Retrieve and print storage shape and stride information for tensors.
• Example:

printf("\nTensor storage shape: %zu", t.storage.count);
print_tensor_inx(t.shape);
print_tensor_inx(t.stride);

---

Code Demonstrations

Base Operations (base_run)

Demonstrates basic tensor creation and manipulation:

• Create a tensor with dimensions (2,3,4).
• Modify specific elements using indexing.
• Permute dimensions for reshaping.

Stack Operations (stack_run)

Illustrates stack-based tensor creation:

• Create tensors with data directly from multi-dimensional arrays.
• Perform permutation on higher-dimensional tensors.

Arithmetic Operations (arith_run)

Showcases element-wise arithmetic operations:

• Generate random tensors within specified ranges.
• Perform addition and multiplication between two tensors.

---

Getting Started

To use this library:

1. Ensure the utilities library is downloaded from [[https://github.com/bipul018/c-utils]]

2. Clone this repo or download the files tensor.h and tensor.c

3. Include the required headers:

#include "tensor.h"

4. Use the provided functions to create and manipulate tensors.

5. Compile tensor.c along with your files.

---