icfpc-2023

codingteam's submission for the ICFP Contest 2023.

News: this year, we won the judges' prize:

Finally, the judges prize was attributed to team Codingteam, whose writeup we really enjoyed reading (and the mix of F# and Fortran in their solution was both surprising and fun to see)

https://icfpcontest2023.github.io/

• July 7th, 12:00 UTC (noon) to July 10th, 12:00 UTC 2023

Team Members

• @Akon32
• @ForNeVeR: link to report in English
• @Minoru: link to report in Russian
• @foxtran
• @gsomix
• @portnov

Latest available rating: we've had 51'st place with a score of 97,433,577,912.

Credentials and Tokens

All the credentials and tokens to access the contest resources are stored in a pinned message in our Telegram chat and in the conference topic in the XMPP chat.

⚠ Please save the access token to file token.txt in the solution directory.

Prerequisites

.NET SDK 7.0 is required. Get it for your platform at https://dotnet.microsoft.com/en-us/download (binary installers are available as well as packages for various Linux distributions).

About F#

• https://thomasbandt.com/fsharp-introduction
• https://learnxinyminutes.com/docs/fsharp/
• https://fsharp.github.io/fsharp-core-docs/
• Just ask @ForNeVeR or @gsomix

Build

$ dotnet build

Run

$ dotnet run --project Icfpc2023 -- [arguments here…]
    # or, alternately:
$ cd Icfpc2023 && dotnet run -- [arguments here…]

Visualizer

$ dotnet run --project Icfpc2023.Visualizer [-- <problemNumber>]
    # or, alternately:
$ cd Icfpc2023.Visualizer && dotnet run [-- <problemNumber>]

Commands

HTTP

Download all problems:

$ dotnet run --project Icfpc2023 -- download all

Upload all solutions from the solutions/ directory:

$ dotnet run --project Icfpc2023 -- upload all

Solution

Solve problem 1 using dummy solver:

$ dotnet run --project Icfpc2023 -- solve 1 dummy

Solve all problems using dummy solver:

$ dotnet run --project Icfpc2023 -- solve all dummy

Solve all problems using non-deterministic solvers and save solutions if they improve upon the existing ones:

$ dotnet run --project Icfpc2023 -- solve all best-nondeterministic --preserve-best

It makes sense to re-run that command from time to time because non-deterministic solvers have a chance of finding a better solution.

Solve problem 1 using all solvers and picking the best solution:

$ dotnet run --project Icfpc2023 -- solve all best

Solve problem 1 using dummy solver and preserve best solution if already exists:

$ dotnet run --project Icfpc2023 -- solve 1 dummy --preserve-best

Calculate score for solution 1 on problem 1:

$ dotnet run --project Icfpc2023 -- score 1

Solution improvement

Improve existing solution of problem 1 using lambda solver:

$ dotnet run --project Icfpc2023 -- improve 1 lambda

Improve existing solution of problem 1 using lambda solver, but preserve best solution:

$ dotnet run --project Icfpc2023 -- improve 1 lambda --preserve-best

Oddities

Lambda score showcase (I dunno, see PR #4 for details):

$ dotnet run --project Icfpc2023 -- lambdaScore

Convert Problem+Solutions to INI:

$ dotnet run --project Icfpc2023 -- convertini <all|problemId>

Import solutions from INI:

$ dotnet run --project Icfpc2023 -- vulpes all # no other options supported, sorry!

Legacy

Download first 3 problems:

$ dotnet run --project Icfpc2023 -- download 3

Run Tests

$ dotnet test

How to Develop

See https://dotnet.microsoft.com/en-us/languages/fsharp/tools for the main options.

Fee free to use editor of choice, but the easiest tools to get started are:

• Visual Studio (Windows-only)

  • Community Edition is free and should be okay for this project
  • Open Infpc2023.sln and then press the Run button

• Visual Studio Code with Ionide extension (cross-platform)

  • Totally free
  • Open the solution folder (the one with the .sln file), install the Ionide extension, and then use the Run and Debug tool window to execute the project

• JetBrains Rider (cross-platform)

  • Commercial
  • Free EAP and trial builds are available and are okay to use for this project (I, @ForNeVeR, use EAP myself)
  • Open the .sln file and then hit the green Run button to get started

Other viable options are Vim, Emacs, and any editor with LSP support via FsAutoComplete.