This is a repo for comparing the capabilities and limitations of ASP solvers on Event Calculus. There are examples for Clingo and s(CASP) including execution logs for different domain sizes.

How to

• directory for each solver contains a number of examples
• all solvers implemented the same examples so that they can be compared
• each example is in its own directory that features a makefile
• use the makefile to run the reasoning (or to see how to do it)
• each example also includes execution logs that show the result, runtime, and memory use

TODO / QUESTIONS for clingo-lpx

• ? Querying values of fluents at an arbitrary timepoint is cumbersome. Currently hacked by introducing check events. Any better way to do this?
• ? Any way to map interpreted values into the answer using #show? (Place things from "Assignment:" directly into "Answer:" in the output)
• ? Any way to say steps with [+-*]e are not allowed? -- might help with the triggered events problem

TODO

• comparison between hybrid-clingo and scasp (maybe examples with growing number of events/steps?)
• ? better benchmarking (currently just very basic one run, no averaging, no timeouts, ...)
• add even more examples?

Notes about the examples

1. ex1-light_on_off
   • The most basic event calculus example. Included as a way to test the implementation of the axioms.
   • The grounding size is not too much of a problem here since there is only one boolean fluent (no parameters) and one non-parametric event.
2. ex2-bank_account_no_fee
   • Discrete fluents and changes. Slightly simplified version of ex4. Included to show the grounding explosion problem.
   • Grounding size becomes a problem fairly quickly since there is a fluent for the amount of money and an event with money as parameter.
3. ex3-falling_object
   • Basic example of continuous change (trajectory + continuous fluent). Included to show inaccuracy issues caused by discretization.
   • Grounding size is a problem.
4. ex4-bank_account
   • Discrete fluents and changes with a triggered event. Preventing repeated trigger becomes tricky when reasoning in continuous time.
   • Requires special techniques or remodeling to avoid zeno-like behavior (infinitely fast response / event occurrence infinitely close to a non-inclusive bound) in continuous time.
5. ex5-light_toggle
   • Very similar to ex1. Uses context dependent event effects via initiates/terminates with a body.
6. e6-adder
   • Testing functional fluents.
7. ex7-pulsing_light
   • Circular trajectories
8. ex8-bouncing_ball
   • Zeno behavior
9. ex9-water_tanks
   • Zeno behavior
10. ex10-blinking_light

• Trivial zeno behavior (infinitely fast frequency)

11. ex11-coin_flip

• Non-deterministic discrete fluent

12. ex12-dice_roll

• Non-deterministic functional fluent