Adding basic fighter implementation for testing

This commit is contained in:
vandomej 2024-03-10 17:41:28 -07:00
parent c44c389fbe
commit d21d0fcd3a
6 changed files with 248 additions and 17 deletions

View file

@ -30,3 +30,4 @@ smol = "2.0.0"
smol-potat = "1.1.2"
num_cpus = "1.16.0"
easy-parallel = "3.3.1"
fann = "0.1.8"

11
gemla/build.rs Normal file
View file

@ -0,0 +1,11 @@
fn main() {
// Replace this with the path to the directory containing `fann.lib`
let lib_dir = "F://vandomej/Downloads/vcpkg/packages/fann_x64-windows/lib";
println!("cargo:rustc-link-search=native={}", lib_dir);
println!("cargo:rustc-link-lib=static=fann");
// Use `dylib=fann` instead of `static=fann` if you're linking dynamically
// If there are any additional directories where the compiler can find header files, you can specify them like this:
// println!("cargo:include={}", path_to_include_directory);
}

3
gemla/config.json Normal file
View file

@ -0,0 +1,3 @@
{
"base_dir": "F:\\\\vandomej\\Projects\\dootcamp-AI-Simulation\\Simulations"
}

View file

@ -1,15 +0,0 @@
[[nodes]]
fabric_addr = "10.0.0.1:9999"
bridge_bind = "10.0.0.1:8888"
mem = "100 GiB"
cpu = 8
# [[nodes]]
# fabric_addr = "10.0.0.2:9999"
# mem = "100 GiB"
# cpu = 16
# [[nodes]]
# fabric_addr = "10.0.0.3:9999"
# mem = "100 GiB"
# cpu = 16

View file

@ -4,6 +4,7 @@ extern crate gemla;
extern crate log;
mod test_state;
mod fighter_nn;
use easy_parallel::Parallel;
use file_linked::constants::data_format::DataFormat;
@ -13,7 +14,7 @@ use gemla::{
};
use smol::{channel, channel::RecvError, future, Executor};
use std::{path::PathBuf, time::Instant};
use test_state::TestState;
use fighter_nn::FighterNN;
use clap::Parser;
#[derive(Parser)]
@ -52,7 +53,7 @@ fn main() -> anyhow::Result<()> {
let args = Args::parse();
// Checking that the first argument <FILE> is a valid file
let mut gemla = log_error(Gemla::<TestState>::new(
let mut gemla = log_error(Gemla::<FighterNN>::new(
&PathBuf::from(args.file),
GemlaConfig {
generations_per_node: 3,

View file

@ -0,0 +1,230 @@
extern crate fann;
use std::{fs::{self, File}, path::PathBuf};
use fann::{ActivationFunc, Fann};
use gemla::{core::genetic_node::GeneticNode, error::Error};
use rand::prelude::*;
use serde::{Deserialize, Serialize};
use serde_json;
use anyhow::Context;
use std::collections::HashMap;
#[derive(Serialize, Deserialize, Debug, Clone)]
struct Config {
base_dir: String,
}
// Here is the folder structure for the FighterNN:
// base_dir/fighter_nn_{fighter_id}/{generation}/{fighter_id}_fighter_nn_{nn_id}.net
// A neural network that utilizes the fann library to save and read nn's from files
// FighterNN contains a list of file locations for the nn's stored, all of which are stored under the same folder which is also contained.
// there is no training happening to the neural networks
// the neural networks are only used to simulate the nn's and to save and read the nn's from files
// Filenames are stored in the format of "{fighter_id}_fighter_nn_{generation}.net".
// The folder name is stored in the format of "fighter_nn_xxxxxx" where xxxxxx is an incrementing number, checking for the highest number and incrementing it by 1
// The main folder contains a subfolder for each generation, containing a population of 10 nn's
#[derive(Serialize, Deserialize, Debug, Clone)]
pub struct FighterNN {
pub id: u64,
pub folder: PathBuf,
pub generation: u64,
// A map of each nn identifier in a generation and their physics score
pub scores: Vec<HashMap<u64, f32>>,
}
impl GeneticNode for FighterNN {
// Check for the highest number of the folder name and increment it by 1
fn initialize() -> Result<Box<Self>, Error> {
// Load the configuration
let config: Config = serde_json::from_reader(File::open("config.json")?)
.with_context(|| format!("Failed to read config"))?;
let base_path = PathBuf::from(config.base_dir);
// Ensure the base directory exists, create it if not
if !base_path.exists() {
fs::create_dir_all(&base_path)?;
}
let mut highest = 0;
let mut folder = base_path.join(format!("fighter_nn_{:06}", highest));
while folder.exists() {
highest += 1;
folder = base_path.join(format!("fighter_nn_{:06}", highest));
}
fs::create_dir(&folder)?;
//Create a new directory for the first generation
let gen_folder = folder.join("0");
fs::create_dir(&gen_folder)?;
// Create the first generation in this folder
for i in 0..10 {
// Filenames are stored in the format of "xxxxxx_fighter_nn_0.net", "xxxxxx_fighter_nn_1.net", etc. Where xxxxxx is the folder name
let nn = gen_folder.join(format!("{:06}_fighter_nn_{}.net", highest, i));
let mut fann = Fann::new(&[10, 10, 10])
.with_context(|| format!("Failed to create nn"))?;
fann.set_activation_func_hidden(ActivationFunc::SigmoidSymmetric);
fann.set_activation_func_output(ActivationFunc::SigmoidSymmetric);
fann.save(&nn)
.with_context(|| format!("Failed to save nn"))?;
}
Ok(Box::new(FighterNN {
id: highest,
folder,
generation: 0,
scores: vec![HashMap::new()],
}))
}
fn simulate(&mut self) -> Result<(), Error> {
// For each nn in the current generation:
for i in 0..10 {
// load the nn
let nn = self.folder.join(format!("{}", self.generation)).join(format!("{:06}_fighter_nn_{}.net", self.id, i));
let fann = Fann::from_file(&nn)
.with_context(|| format!("Failed to load nn"))?;
// Simulate the nn against the random nn
let mut score = 0.0;
// Using the same original nn, repeat the simulation with 5 random nn's from the current generation
for _ in 0..5 {
let random_nn = self.folder.join(format!("{}", self.generation)).join(format!("{:06}_fighter_nn_{}.net", self.id, thread_rng().gen_range(0..10)));
let random_fann = Fann::from_file(&random_nn)
.with_context(|| format!("Failed to load random nn"))?;
let inputs: Vec<f32> = (0..10).map(|_| thread_rng().gen_range(-1.0..1.0)).collect();
let outputs = fann.run(&inputs)
.with_context(|| format!("Failed to run nn"))?;
let random_outputs = random_fann.run(&inputs)
.with_context(|| format!("Failed to run random nn"))?;
// Average the difference between the outputs of the nn and random_nn and add the result to score
let mut round_score = 0.0;
for (o, r) in outputs.iter().zip(random_outputs.iter()) {
round_score += o - r;
}
score += round_score / fann.get_num_output() as f32;
}
score /= 5.0;
self.scores[self.generation as usize].insert(i, score);
}
Ok(())
}
fn mutate(&mut self) -> Result<(), Error> {
// Create the new generation folder
let new_gen_folder = self.folder.join(format!("{}", self.generation + 1));
fs::create_dir(&new_gen_folder)?;
// Remove the 5 nn's with the lowest scores
let mut sorted_scores: Vec<_> = self.scores[self.generation as usize].iter().collect();
sorted_scores.sort_by(|a, b| a.1.partial_cmp(b.1).unwrap());
let to_keep = sorted_scores[5..].iter().map(|(k, _)| *k).collect::<Vec<_>>();
// Save the remaining 5 nn's to the new generation folder
for i in 0..5 {
let nn_id = to_keep[i];
let nn = self.folder.join(format!("{}", self.generation)).join(format!("{:06}_fighter_nn_{}.net", self.id, nn_id));
let new_nn = new_gen_folder.join(format!("{:06}_fighter_nn_{}.net", self.id, i));
fs::copy(&nn, &new_nn)?;
}
// Take the remaining 5 nn's and create 5 new nn's by the following:
for i in 0..5 {
let nn_id = to_keep[i];
let nn = self.folder.join(format!("{}", self.generation)).join(format!("{:06}_fighter_nn_{}.net", self.id, nn_id));
let mut fann = Fann::from_file(&nn)
.with_context(|| format!("Failed to load nn"))?;
// For each weight in the 5 new nn's there is a 20% chance of a minor mutation (a random number between -0.1 and 0.1 is added to the weight)
// And a 5% chance of a major mutation (a random number between -0.3 and 0.3 is added to the weight)
let mut connections = fann.get_connections(); // Vector of connections
for c in &mut connections {
if thread_rng().gen_range(0..100) < 20 {
c.weight += thread_rng().gen_range(-0.1..0.1);
} else if thread_rng().gen_range(0..100) < 5 {
c.weight += thread_rng().gen_range(-0.3..0.3);
}
}
fann.set_connections(&connections);
// Save the new nn's to the new generation folder
let new_nn = new_gen_folder.join(format!("{:06}_fighter_nn_{}.net", self.id, i + 5));
fann.save(&new_nn)
.with_context(|| format!("Failed to save nn"))?;
}
self.generation += 1;
self.scores.push(HashMap::new());
Ok(())
}
fn merge(left: &FighterNN, right: &FighterNN) -> Result<Box<FighterNN>, Error> {
// Find next highest
// Load the configuration
let config: Config = serde_json::from_reader(File::open("config.json")?)
.with_context(|| format!("Failed to read config"))?;
let base_path = PathBuf::from(config.base_dir);
// Ensure the base directory exists, create it if not
if !base_path.exists() {
fs::create_dir_all(&base_path)?;
}
let mut highest = 0;
let mut folder = base_path.join(format!("fighter_nn_{:06}", highest));
while folder.exists() {
highest += 1;
folder = base_path.join(format!("fighter_nn_{:06}", highest));
}
fs::create_dir(&folder)?;
//Create a new directory for the first generation
let gen_folder = folder.join("0");
fs::create_dir(&gen_folder)?;
// Take the 5 nn's with the highest scores from the left nn's and save them to the new fighter folder
let mut sorted_scores: Vec<_> = left.scores[left.generation as usize].iter().collect();
sorted_scores.sort_by(|a, b| a.1.partial_cmp(b.1).unwrap());
let mut remaining = sorted_scores[5..].iter().map(|(k, _)| *k).collect::<Vec<_>>();
for i in 0..5 {
let nn = left.folder.join(format!("{}", left.generation)).join(format!("{:06}_fighter_nn_{}.net", left.id, remaining.pop().unwrap()));
let new_nn = folder.join(format!("0")).join(format!("{:06}_fighter_nn_{}.net", highest, i));
trace!("From: {:?}, To: {:?}", &nn, &new_nn);
fs::copy(&nn, &new_nn)
.with_context(|| format!("Failed to copy left nn"))?;
}
// Take the 5 nn's with the highest scores from the right nn's and save them to the new fighter folder
sorted_scores = right.scores[right.generation as usize].iter().collect();
sorted_scores.sort_by(|a, b| a.1.partial_cmp(b.1).unwrap());
remaining = sorted_scores[5..].iter().map(|(k, _)| *k).collect::<Vec<_>>();
for i in 5..10 {
let nn = right.folder.join(format!("{}", right.generation)).join(format!("{:06}_fighter_nn_{}.net", right.id, remaining.pop().unwrap()));
let new_nn = folder.join(format!("0")).join(format!("{:06}_fighter_nn_{}.net", highest, i));
trace!("From: {:?}, To: {:?}", &nn, &new_nn);
fs::copy(&nn, &new_nn)
.with_context(|| format!("Failed to copy right nn"))?;
}
Ok(Box::new(FighterNN {
id: highest,
folder,
generation: 0,
scores: vec![HashMap::new()],
}))
}
}