PsiML++

PsiML++ is a modern C++ machine learning library designed for high-performance computations, providing core utilities, linear algebra operations, machine learning algorithms, preprocessing tools, and vision functionalities. The library is structured for modular development, testing, and easy integration into C++ projects.

---

Table of Contents

1. Project Overview
2. Directory Structure
3. Features
4. Dependencies
5. Building the Project
6. Running Tests
7. Modules Overview
8. Contributing
9. License

---

Project Overview

PsiML++ is a versatile C++ library aimed at researchers, developers, and engineers who need:

• Fast and efficient matrix/tensor operations.
• Built-in machine learning algorithms like linear regression, logistic regression, PCA, KMeans, and SVM.
• Preprocessing utilities for dataset normalization, encoding, and scaling.
• Vision tools for image I/O, processing, and drawing.
• Utilities for logging, profiling, and serialization.

The library leverages modern C++ features and is designed to be cross-platform with minimal dependencies.

---

Directory Structure

PsiML++
│   PsiML++.slnx                  # Visual Studio 2026 solution file
│   .gitignore
│   .gitattributes
│
├── PsiML++/                       # Main source folder
│   ├── include/                   # Header files
│   │   ├── core/                  # Core utilities (logging, memory, device)
│   │   ├── math/                  # Linear algebra and math utilities
│   │   │   ├── linalg/            # BLAS, solvers, decomposition
│   │   │   └── ops/               # Arithmetic, broadcasting, reduction
│   │   ├── ml/                    # Machine learning algorithms
│   │   │   ├── algorithms/        # KMeans, Linear/Logistic Regression, PCA, SVM
│   │   │   ├── optimizers/        # Gradient Descent, Momentum, SGD
│   │   │   └── preprocessing/     # Encoders, normalizers, scalars
│   │   ├── utils/                  # Data loaders, file IO, model IO, profiling
│   │   └── vision/                 # Image processing, drawing
│   │
│   ├── src/                        # Source files (.cpp)
│   │   ├── core/
│   │   ├── math/
│   │   │   └── linalg/
│   │   └── vision/
│   │
│   ├── dependencies/               # Third-party libraries (if any)
│   ├── tests/                      # Unit tests and test data
│   │   ├── bin/                    # Compiled test binaries
│   │   ├── data/                   # Sample CSV datasets
│   │   └── *.cpp                   # Individual test files
│   └── x64/                        # Build output (Debug/Release)

---

Features

Core Utilities

• Device management, memory management, custom exception handling, logging, timers, and configuration management.

Math & Linear Algebra

• Tensor, matrix, vector types
• Linear algebra operations: BLAS routines, decompositions, eigenvalues, solvers, statistics
• Element-wise operations: arithmetic, broadcasting, reduction, statistics

Machine Learning

• Algorithms: Linear Regression, Logistic Regression, PCA, KMeans, SVM
• Optimizers: Gradient Descent, Momentum, SGD
• Preprocessing: Normalizers, Encoders, Scalar transformations
• Dataset handling and metrics evaluation

Vision & Image Processing

• Image I/O, drawing primitives, and processing operations

Utilities

• File I/O, Model serialization/deserialization, data loading, string utilities, profiler

---

Dependencies

• Visual Studio 2026
• C++17 or higher
• Linking OpenCV (for image processing in vision module)

---

Building the Project

1. Open PsiML++.slnx in Visual Studio 2026.
2. Select your desired build configuration: Debug or Release.
3. Build the solution using Build → Build Solution.
4. The output libraries and binaries will be generated under x64/Debug or x64/Release.

Notes:

• All source files are organized into modules (core, math, ml, vision) for modular compilation.
• Dependencies (like OpenCV) should be linked via the Project Properties → VC++ Directories → Include/Library Directories.

---

Running Tests

Tests are located in PsiML++/tests. You can run them individually or collectively:

Using Visual Studio:

• Open the test project in the solution and run tests from Test Explorer.

Using Batch Script:

cd PsiML++\tests
compile_and_run.bat

Test Coverage

• test_arithmetic.cpp, test_blas.cpp, test_decomposition.cpp, test_ml.cpp, test_vision.cpp, and more.
• Test datasets are included under tests/data (iris_binary.csv, housing.csv, clustering.csv).

---

Modules Overview

Core

• config.h, device.h, memory.h, logging.h, exception.h
• Provides basic utilities required for ML computation.

Math

• matrix.h, tensor.h, vector.h, random.h
• Linear algebra (blas.h, solvers.h, decomposition.h) and operations (arithmetic.h, broadcasting.h, reduction.h)

ML

• Algorithms: kmeans.h, linear_regression.h, logistic_regression.h, pca.h, svm.h
• Optimizers: gradient_descent.h, momentum.h, sgd.h
• Preprocessing: encoder.h, normalizer.h, scalar.h
• Utilities: dataset.h, metrics.h, model.h

Vision

• Image manipulation, drawing, and processing via image.h, drawing.h, image_io.h, image_processing.h

Utilities

• data_loader.h, file_io.h, model_io.h, profiler.h, serialization.h, string_utils.h, timer.h

---

Contributing

• Fork the repository and create a feature branch.
• Follow the folder structure when adding new modules.
• Include corresponding tests for new functionalities.
• Submit pull requests with detailed descriptions.

---

License

MIT License