You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In this case, it's "simple" for someone with a private algorithm to plug their algorithm into a new Docker container into a
Kubernetes or local simulation.
For example, someone is trying to swap out the state estimator with their own algorithm, in order to benchmark their algorithm. Everything would run entirely locally, entirely on Kubernetes, entirely on Docker, etc.
The algorithm communicates with the HELICS broker directly in a standard way.
An algorithm developer wants to provide the use of an algorithm to external simulations without exposing their code. As an example, an algorithm developer may want to provide limited use for testing without compromising their intellectual property.
The simulation execution needs to contact the algorithm to start a simulation.
There needs to be a container published to Docker Hub which behaves identically to an open algorithm, but calls out to an external algorithm server for core computation.
An external algorithm server manages multiple simulations at once. Each "control" container contacts the server to spin up a new instance for that simulation, and results are communicated to the control container. All communication takes place through a REST interface.
Additional burden is placed on algorithm developers to provide this "split" setup. It will probably never be "easy".
The text was updated successfully, but these errors were encountered:
Simulation Executor = Algorithm Developer (Private/Private Use-Case) ("Easy" case)
In this case, it's "simple" for someone with a private algorithm to plug their algorithm into a new Docker container into a
Kubernetes or local simulation.
For example, someone is trying to swap out the state estimator with their own algorithm, in order to benchmark their algorithm. Everything would run entirely locally, entirely on Kubernetes, entirely on Docker, etc.
Simulation Executor != Algorithm Developer (Public/Private Use-Case)
An algorithm developer wants to provide the use of an algorithm to external simulations without exposing their code. As an example, an algorithm developer may want to provide limited use for testing without compromising their intellectual property.
Additional burden is placed on algorithm developers to provide this "split" setup. It will probably never be "easy".
The text was updated successfully, but these errors were encountered: