"""Train a model of grid transitions and evaluate policies on disjoint layouts.""" import argparse import hashlib import json import random from collections import deque from pathlib import Path import numpy as np from next_reward import MOVES, NextReward ORIGINAL=(".....",".X...",".X...","...X.","....G") def key(board): return hashlib.sha256("\n".join(board).encode()).hexdigest() def generate(rng,n,wall=.20): cells=rng.choices(".#X",weights=[1-wall-.1,wall,.1],k=n*n) cells[rng.randrange(n*n)]="G" return tuple("".join(cells[i:i+n]) for i in range(0,n*n,n)) def advance(b,pos,a): y,x=pos r,c=y+MOVES[a][0],x+MOVES[a][1] if not (0<=r