herrmanno/informaticup-2023

InformatiCup 2023

Solver for the 2023 InformatiCup challenge „Profit“

Install / Run

Build

cargo build --release [--features stats]

If the stats feature is activated, the solver will print the number of calculated solutions per second.

Note: For competition, the solver shall be run as docker container.

Docker build

docker build -t .

Docker run

docker run -i [OPTIONS] <

Packages

The project consists of multiple packages:

Solver

Solver for puzzles

Example

target/release/solver
    --time [runtime in seconds]
    --cores [number of threads to use]
    [--print] # prints the final solution to stdout
    [--stats] # prints evaluation stats (score and turn, when score was achieved) to stdout
    < some_task.json

Quality Assurance

Test runner for a fixed set of tasks.

Example

target/release/qa

Printer

Binary to read task and solution files and print resulting map

Example

target/release/printer -- < some_task.json

Simulator

Binary for running a simulation from a task/solution file

Example

target/release/simulator -- < some_task.json

Model

Model instances for task/solution files and basic building objects

Test

Run benchmarks

cargo bench --bench benchmarks

Profile

To create profiling data:

cargo build --profile=profiling
valgrind --tool=callgrind ./target/profiling/solver --time x --cores y < inputs/003.task.json

To view profiling data (w/ qcachegrind)

qcachegrind callgrind.out.xxxx

Input formats

Task file

A JSON object that contains the tasks basic information and given objects (may also include non-landscape objects. In that case the task is split into task description and solution).