osqp 1.1.0

Dart bindings for the OSQP quadratic programming solver.

package:osqp

Dart bindings for the OSQP (Operator Splitting Quadratic Program) solver.

OSQP is a numerical optimization package for solving convex quadratic programs. Quadratic programming is a superset of linear programming, so this solver can also handle linear programming problems.

Loads of real world optimisation problems can be formulated as linear or quadratic programming problems. For example, I wrote this package so that I could use OSQP to optimise my home solar panel and battery.

Features

• High-level Dart API: Idiomatic wrappers around OSQP's C API. For example, the C OSQPSolver is wrapped by the Dart Solver, and OSQPCscMatrix is wrapped by Matrix.
• Automatic Memory Management: Memory is freed automatically via NativeFinalizers.
• Cross-Platform Build Hooks: Automated native compilation using Dart build hooks on Linux, MacOS, and Windows (probably works on iOS and Android too, but I haven't tested it).
• Raw FFI Access: You can also use the FFI bindings directly if you need low level access.

Usage

See example/osqp_example.dart for a complete working example. For more information, check out the OSQP documentation.

Building

Generally it's not necessary to build OSQP for ordinary usage. The published package contains prebuilt binaries for all the common targets. So no CMake or C compiler is required.

If a prebuilt binary is not available for your target platform, or if you want to compile OSQP from source locally, the Dart build hook automatically falls back to building the native OSQP C library using CMake. Local builds require:

• CMake (version 3.18 or newer)
• A C/C++ compiler toolchain (e.g., GCC, Clang, or MSVC)

You can force a local source build in your app's pubspec.yaml via hook user defines:

hooks:
  user_defines:
    osqp:
      local_build: true