2 releases (1 stable)
1.0.0 | Apr 28, 2020 |
---|---|
0.1.0 | Nov 23, 2019 |
#443 in Games
145KB
1.5K
SLoC
2048-rs
This is a Rust implementation of the famous 2048 game which runs in the terminal.
In addition to the standard user inputs, an AI can be used to perform the moves. This AI leverages the expectiminimax algorithm to recommend the best next move at each step.
Statistics
Here are the statistics of the AI with its default parameters:
max tile reached | frequency |
---|---|
2048 | 100 % |
4096 | 100 % |
8192 | 96 % |
16384 | 70 % |
32768 | 10 % |
In particular, these statistics correspond to a minimum branch probability of 0.001
.
Decreasing this value would lead to better performance, as more branches would be explored, but this would also take more time.
Some other hardcoded parameters can be tweaked in order to further improve the algorithm. I have not performed an exhaustive grid search, thus the parameter set is probably sub-optimal.
Installation with Cargo
cargo install play-2048
Then, in order to play:
play-2048
Installation from source
> git clone https://github.com/adrienball/2048-rs.git
> cd 2048-rs
> cargo run --release
Usage
You can change the probability of drawing a 4 tile:
> play-2048 --proba-4 0.5
Or adjust the minimum branch probability of the expectiminimax search:
> play-2048 --min-branch-proba 0.0001
To get the list of available options, simply run:
> play-2048 --help
License
Apache 2.0/MIT
All original work licensed under either of
- Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
- MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT) at your option.
Contribution
Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.
Dependencies
~3–4MB
~67K SLoC