Artifacts for "Inductive Synthesis of Recursive Functional Programs with Trace-based Pruning"

Build

$ git clone https://github.com/pslhy/trio_artifacts.git
$ cd trio_artifacts
$ ./build
$ . setenv # or ($ bash setenv)
$ make

Running All the Experiments

$ python3 artifact.py  [ io | ref | ablation ] [--timeout <sec> (default: 120)]

To run all the IO Example Benchmarks, run the following command:

$ python3 artifact.py io

To run all the Reference Implementation Benchmarks, run the following command:

$ python3 artifact.py ref

The following command will run all the Ablations Study section:

$ python3 artifact.py ablation

Effectiveness of Trio with IO Specifications (Table 2 in the paper)

After running the above commands you can see the results of Table 2 in the paper by running the following command:

$ python3 artifact.py --print_result 1

Effectiveness of Trio with Ref Specifications (Table 3 in the paper)

You can reproduce the results of Table 3 in the paper by running the following command:

$ python3 artifact.py --print_result 2

Effectiveness of the block-based pruning and library sampling techniques

To run following command, you can see the summary of ablation study:

$ python3 artifact.py --print_result 3

Running Trio for a single benchmark

You can run Trio to solve single benchmark as follows:

$ burst/BurstCmdLine.exe -use-trio [benchmark file]

and if you run with option of Trio,

$ burst/BurstCmdLine.exe -use-trio -trio-options “<trio options>” [benchmark file]

For example, to solve the single benchmark in benchmarks/io/nat_mul.mls

$ burst/BurstCmdLine.exe -print-data -use-trio benchmarks/io/nat_mul.mls

will print the following output:

fix (f : nat * nat -> nat) =
  fun (x:nat * nat) ->
    match x . 1 with
      | O _ -> x . 1
      | S _ -> add (f (Un_S (x . 1), x . 0)) (x . 0)
Size: 27
Iter: 0

At 1-5 lines show solution, and other lines shows solution size and num of iter times by -print-data option.

Acknowledgement

This artifact is based on the artifact for Burst