Welcome to slap’s documentation!

Why a library for simple linear algebra protocols? Sometimes you just need something simple. Something you can understand. Something you can debug. Something that just works.

When it comes to linear algebra in C, the go-to options are often huge dependencies, difficult to install, hard to learn, and difficult to read. (I’m looking at you, BLAS, like who would guess that potrf is a Cholesky decomposition?)

Priorities

While most linear algebra packages prioritize speed and performance (often at the cost of everything else), this library prioritizes the following:

  1. Simplicity: a new-comer should be able to start using it effectively within 10-30 minutes.

  2. Readability: code using this library should be easy to read and follow

  3. Debugging: it should be easy to step through and quickly find and fix errors.

  4. Lightweight: it should be able to be used on almost any platform, especially those with limited compute

Key Design Principles

The library uses the following design principles:

  1. Matrices are stored in a column-major format

  2. All matrices are a single data type (double)

  3. Matrices have a row stride of 1 (all elements in a column are adjacent in memory)

  4. A matrix is just a pointer with some metadata

  5. A matrix (as just a pointer with some metadata) is passed by value

  6. Memory is owned by the user, and is never allocated by the library (except by the explicit NewMatrix and FreeMatrix commands)

  7. Error checking is primarily the responsibility of the user, since extensive error checking is computationally expensive

(especially for low-level functions like indexing)

Table of Contents

Indices and tables