Skip to content

Latest commit

 

History

History
174 lines (140 loc) · 5.93 KB

README.md

File metadata and controls

174 lines (140 loc) · 5.93 KB
Raw

Bi-criteria-Search-Algorithms

A software project at the SPbU.

Description

This project contains algorithms implementations that solves Bi-criteria pathfinding problem.

Getting started

For project build you should have compiler on C++17 standart. For example, you can install it with

sudo apt install build-essential

This project should be built with CMake. For example, you can install it with

sudo apt install cmake

Also you will need Boost library. For example, you can install it with

sudo apt install libboost-all-dev

Also you will need Python3.8

For Python you will need tqdm and Pil. You can install all dependencies from requirements.txt with

pip install -r requirements.txt

Installing

To download current repository to your local machine use

git clone https://github.com/bagrorg/Bi-criteria-Search-Algorithms

To build the project follow the guide

cd Bi-criteria-Search-Algorithms/
mkdir build
cd build
cmake ..
make -j nproc

Arguments

You can manage settings of execution using arguments.

  • algo - wich algorithm to use. Now supported - BOA*, BOA*-epsilon and PP-A* (to use BOA*-epsilon just use BOA* and pass epsilons)
  • report_path - where to save execution description
  • graph_path - path to graph description file
  • iterations - how much executions to provide
  • eps_dist, eps_time - epsilons for PP-A* and BOA*-epsilon algorithms for distance cost and time cost
  • start_id, goal_id - indexes of start and goal vertexes
  • history_path - where to save history of OPEN and CLOSEd
  • h_time_path, h_dist_path - path to heuristic files (e.g. dijkstra results)

Also you can run explanation with

./main --help

Input and output

3 sepparate input graph files

For input there must be 3 files:

  • file with time cost
  • file with distance cost
  • file with coordinates

Indexing of vertices must starts from 1! Format:

  • file with time/distance cost Lines must be like
c {comments}
p sp {count of vertices} {count of edges}

a {vertex from} {vertex to} {time/distance cost}
  • file with coordinates Lines must be like
c {comments}

v {vertex index} {x} {y}

After this, you can merge this files to one with script mergeGraphFiles.py. To check how to use this script see Scripts. This file you will need for execution

1 input graph file

  • file with time, distance and coordinates Lines must be
c {comments}
p sp {count of vertices} {count of edges}
v {vertex index} {x} {y}
a {vertex from} {vertex to} {time cost} {distance cost}

Indexing of vertices must starts from 0!

You can find examples here. Also you can find .gr/.co files on 9th DIMACS Implementation Challenge: ShortestPath. Check Sources for more information.

Output files

  • history file - contains all OPEN or solutions updates while an execution Format:
e {num of epoch (iteration)}

n {nodes that was added on this iteration}

s {solution that was added on this iteration}

  • gif - gif file of working process

  • solutions - Paretto-optimal frontier Format

s {dist} {time}
p {path}

Examples here

Misc

  • heuristics (from dijkstra.py) Format
{vertex id} {heuristic}

Scripts

  • mergeGraphFiles.py {time path} {dist path} {coord path} {output path} - merging 3 graph files at {time path}, {dist path} and {coord path} to one at {output path}. Indexing must starts from 1.

  • dijkstra.py {input graph path} {output path} {vertex id} - counting dijkstra results (for heuristics) using graph at {input graph path} for {vertex id} vertex and saving it at {output path} in format

{vertex id} {result}

Indexing must starts from 0.

  • renderHistory.py {history path} {graph path} {output path} {start id} {goal id} {duration} {max frames} {width} {height} - making a gif from hastory file at {history path} using graph at {graph path} and saving it at {output path}. History is for path starting at {start id} and ending at {goal id}. {duration}, {max frames}, {height} and {width} - settings for gif output. Can be -1 for default. Indexing must starts from 0.

  • serviceForDijkstra.py - contains interfaces for dijkstra.py

  • run.py - main script that runs program on input graph

    • {--time_graph_path}, {--dist_graph_path}, {--coords_graph_path} - paths to input files
    • {--output_graph_path}, {--output_graph_path}, {--dist_heuristic_path} - paths to output files
    • {--start_id}, {--goal_id} - vertexes indexes
    • {--app_path} - path to c++ built program
    • {--algo_name} - algorithm (e.g. BOA*)
    • {--history_path} - path to history output (optional)
    • {--eps_time}, {--eps_dist} - epsilons (optional)
    • {--report_path} -- report path (optional) Example
python3.9 run.py --time_graph_path ../Data/USA-road-t.NY.gr --dist_graph_path ../Data/USA-road-d.NY.gr --coords_graph_path ../Data/USA-road-d.NY.co --output_graph_path ../Data/Res.gr --output_graph_path ../Data/timeHeur.txt --dist_heuristic_path ../Data/distHeur.txt --start_id 31231 --goal_id 0 --app_path ../build/main --algo_name BOA* --eps_time 0 --eps_dist 0 --report_path ../Data/report.csv

Sources

  • Hernández Ulloa, C., Yeoh, W., Baier, J. A., Zhang, H., Suazo, L., & Koenig, S. (2020). A Simple and Fast Bi-Objective Search Algorithm. URL
  • Goldin, B., & Salzman, O. (2021). Approximate Bi-Criteria Search by Efficient Representation of Subsets of the Pareto-Optimal Frontier. URL
  • 9th DIMACS Implementation Challenge: ShortestPath. URL

Participants

Mentor

Yakovlev Konstantin Sergeevich

Students

Kulikov Daniil

Sadykov Rustam