Dense and sparse double-precision BLAS Levels 1–3 for Kotlin Multiplatform, using DoubleArray storage. Every supported operation works without a native library. JVM SIMD and optional native acceleration are selected automatically. Factorizations and general solvers belong to consumers.

Add the dependency to your JVM or commonMain source set:

implementation("com.eignex:koblas:<version>")

Supported: JDK 17+, Android API 24+, Linux x64/arm64, Windows x64, macOS arm64, and iOS arm64 devices and simulators. On desktop JDK 25+, enable SIMD with --add-modules=jdk.incubator.vector. See engine loading and native libraries for optional acceleration and setup.

import com.eignex.koblas.*

val a = DenseMatrix.ofRows(arrayOf(
    doubleArrayOf(2.0, 1.0),
    doubleArrayOf(1.0, 3.0),
))
val x = DenseVector.of(doubleArrayOf(3.0, 5.0))
val y = a * x

Here a * x is a matrix-vector product. Operators return new results; calls ending in Into write into existing arrays. See the usage guide for more examples and reusing temporary storage.

• Engines and native libraries: loading diagram, function coverage, thresholds and measured gains.
• Usage: vector and matrix operations, reusable arrays, views, and Java and mobile setup.
• API overview: dense, sparse and raw array operations.
• Benchmarks: reproducible comparisons and measurement reports.