An extremely fast and flexible C++ simulation engine and differential equation solver.
Built from the ground up as a major advancement of Ascent-beta
See the Ascent Wiki for more examples and help
Study the modular-spring-damper example to learn the basics of solving matrix free systems of ordinary differential equations in an object-oriented manner
Ascent's integration algorithms are designed for speed, and outperform boost's odeint in both Debug and Release
Integration algorithms are automatically vectorized when using std::vector
Ascent solvers conform to the odeint system syntax, letting you run odeint system without changes (and with faster results!)
odeint solvers can also be used to run Ascent simulations, providing all flexibility of the odeint engine with Ascent's simulation framework
Ascent can solve complex, dynamic systems of differential equations in a modular, object-oriented manner
Easily multi-thread systems and change integrators on the fly
- Header Only
- Automatic Vectorization: Ascent conforms to vectorization standards (such as Intel's)
- Free for open source and commercial applications (Apache License)
- Modular (Optional): solve systems in an object-oriented manner
- Variable Tracking: Optimized recording of variable time history
- Asynchronous Sampling and Event Scheduling
- Multiple Integration Algorithms (In Progress): adaptive steppers, predictor-correctors, etc.
- Use boost's odeint library as the numerical integration engine
- Scripting Interface: Optional ChaiScript interface to easily script simulations
- Aerospace, multi-body physics, chemical reactions, economics, circuits, and much more
- As a game engine for synchronization and physics
- Agent-based simulations
- Complex systems of differential equations
- State-space modeling
- Control algorithms (e.g. robotics)
- C++17 compliant compiler
- ChaiScript Embedded Scripting Language
Ascent-beta still remains a state of the art simulation engine, far surpassing the capabilities of many modern engines/ode solvers.
- Header only
- Much, much faster. With a simple spring-mass-damper simulation, the current version is twelve-times faster than the beta version.
- Cleaner code: Ascent doesn't require as many specialized containers, avoids pointers, and takes advantage of more core C++. This makes it faster to write simulations, the code is more comprehensible, and it is easier to debug.
- State-space modeling: Ascent allows state space modeling and allows it to be integrated with modular design.
- Module/Simulation abstraction: Modules are only simulation specific if they directly handle integration states. This means modules can easily be used across simulations even while running.
- Easier, faster, and more powerful scripting: Simulation loops can now be scripted in Ascent. Scripts are also easier to move to C++ if the user wants to compile simulation designs.
- More straightforward multi-threading
- This new version completely separates concepts of simulators, integrators, systems, modules, recorders, and more.
- Integration algorithms have reduced memory footprints (RK4 uses less than half the memory as before)
About odeint
odeint is a C++ ordinary differential equation solver that is part of the boost library.
Ascent was partly inspired by the design of odeint, but Ascent offers better performance where comparisons can be made, this is especially true for solving object-oriented systems.
odeint offers various state types and solvers that Ascent integration algorithms currently do not support, so the odeint solvers are a viable option as the integration algorithm beneath an Ascent simulation.