Matr is an attempt to bring together safety, flexibility and simplicity for matrix calculations.
- Safety through typed shapes and immutable data
- Flexibility through loose coupling between interface, data and operations
- Simplicity through an easily accessible API and "batteries included" default implementations, both with zero dependencies
Add the following dependency to your SBT build configuration:
libraryDependencies += "io.github.dieproht" %% "matr-bundle" % "<latest version>" // Bring API interfaces in scope
import matr.{Matrix, MatrixFactory}
// Bring bundled implementations in scope
import matr.MatrBundle.given
// Create a Matrix from tuples
val a: Matrix[3, 4, Int] =
MatrixFactory[3, 4, Int].fromTuple(
(0, 8, 15, 0),
(4, 7, 1, 1),
(1, 2, 3, 4)
)
// Create a Matrix of ones
val b: Matrix[3, 4, Int] = MatrixFactory[3, 4, Int].ones
// Add two matrices
val c: Matrix[3, 4, Int] = a + b
// Create a Matrix of random numbers
val d: Matrix[4, 2, Int] =
MatrixFactory[4, 2, Int].tabulate((_, _) => scala.util.Random.nextInt(20))
// Calculate the dot product of two Matrices
val e: Matrix[3, 2, Int] = c dot d
// Transpose a Matrix
val f: Matrix[2, 3, Int] = e.transpose
// Pretty print a Matrix
println(f.mkString)
// ... it outputs something like this:
// ⎛ 73, 77, 116⎞
// ⎝430, 253, 179⎠Matr checks the shape of a Matrix at compile-time, thus the following code does not compile:
// Element (2, 3) is missing when creating the Matrix
val ae =
MatrixFactory[3, 4, Int].fromTuple(
(0, 8, 15, 0),
(4, 7, 1, 1),
(1, 2, 3)
)
// Adding two Matrices with different dimensions is not possible
val be = MatrixFactory[2, 5, Int].ones
val ce = a + be
// Column dimension of left-hande side and row dimension of right-hand side
// must be equal when calculating dot product
val de = MatrixFactory[1, 2, Int].tabulate((_, _) => scala.util.Random.nextInt(20))
val ee = c dot deThis project is currently in an early phase. If you are looking for a mature tool for matrix calculations in Scala, you should definitely consider Breeze!
All code is available to you under the Apache 2 license, available at https://opensource.org/licenses/Apache-2.0.
Copyright Karl F Walkow, 2020 - 2021.