app.py

import random
import numpy as np
import streamlit as st

# Default parameter ranges
default_param_ranges = {
    'beef': (400, 600),
    'cornflour': (80, 120),
    'garlic': (2, 4),
    'cilantro': (10, 20),
    'salt': (0.5, 1.5),
    'pepper': (0.25, 0.75),
    'baking_powder': (0.25, 0.75)
}

# Explanations for each parameter
param_explanations = {
    'beef': 'Amount of beef in grams.',
    'cornflour': 'Amount of cornflour in grams.',
    'garlic': 'Number of garlic cloves.',
    'cilantro': 'Amount of cilantro in grams.',
    'salt': 'Amount of salt in teaspoons.',
    'pepper': 'Amount of pepper in teaspoons.',
    'baking_powder': 'Amount of baking powder in teaspoons.'
}

def generate_recipe(param_ranges):
    return {k: random.uniform(v[0], v[1]) for k, v in param_ranges.items()}

def simulate_mixing(recipe, mix_factor):
    beef = recipe['beef']
    cornflour = recipe['cornflour']
    # Simulate mixing by dividing beef by cornflour and adding mix factor
    
    # If cornflour is 0, return 0
    if cornflour == 0:
        return 0
    mix_factor = (beef / cornflour) + mix_factor

    # If mix factor is greater than 1, return 1
    if mix_factor > 1:
        return 1

    return mix_factor

def evaluate_squishiness(recipe):
    beef = recipe['beef']
    cornflour = recipe['cornflour']
    baking_powder = recipe['baking_powder']
    squishiness = (beef / cornflour) + (baking_powder * 10)
    return squishiness

def evaluate_meat_taste(recipe, taste_weights):
    beef = recipe['beef']
    garlic = recipe['garlic']
    cilantro = recipe['cilantro']
    meat_taste = (beef * taste_weights['beef']) + (garlic * taste_weights['garlic']) + (cilantro * taste_weights['cilantro'])
    return meat_taste

def evaluate_consistency(recipe, mix_factor):
    beef = recipe['beef']
    cornflour = recipe['cornflour']
    consistency = (beef / cornflour) * mix_factor
    return consistency

def evaluate_fitness(recipe, weights, mix_factor, taste_weights):
    squishiness = evaluate_squishiness(recipe)
    meat_taste = evaluate_meat_taste(recipe, taste_weights)
    consistency = evaluate_consistency(recipe, mix_factor)
    fitness = (squishiness * weights['squishiness']) + (meat_taste * weights['meat_taste']) + (consistency * weights['consistency'])
    return fitness

def genetic_algorithm(param_ranges, weights, pop_size=100, generations=50, mutation_rate=0.1, mix_factor=0.1, taste_weights=None):
    population = [generate_recipe(param_ranges) for _ in range(pop_size)]
    for gen in range(generations):
        fitness_scores = [evaluate_fitness(recipe, weights, mix_factor, taste_weights) for recipe in population]
        parents = select_parents(population, fitness_scores, pop_size // 2)
        offspring = []
        for i in range(0, len(parents), 2):
            if i + 1 < len(parents):
                child = crossover(parents[i], parents[i + 1])
                child = mutate(child, mutation_rate, param_ranges)
                offspring.append(child)
        population = parents + offspring
    best_recipe = select_parents(population, fitness_scores, 1)[0]
    return best_recipe

def select_parents(population, fitness_scores, num_parents):
    parents_indices = np.argsort(fitness_scores)[-num_parents:]
    return [population[i] for i in parents_indices]

def crossover(parent1, parent2):
    child = {}
    for param in parent1.keys():
        child[param] = (parent1[param] + parent2[param]) / 2
    return child

def mutate(recipe, mutation_rate, param_ranges):
    for param in recipe.keys():
        if random.random() < mutation_rate:
            recipe[param] += random.uniform(-0.1, 0.1) * recipe[param]
            recipe[param] = max(param_ranges[param][0], min(recipe[param], param_ranges[param][1]))
    return recipe

# Streamlit app
st.title('🧆 Meatball Recipe Simulator')

# Custom parameter ranges
st.sidebar.header('Custom Parameter Ranges')
param_ranges = {}
for param, (low, high) in default_param_ranges.items():
    st.sidebar.markdown(f'**{param.capitalize()}**')
    st.sidebar.write(param_explanations[param])
    low_val = st.sidebar.number_input(f'{param.capitalize()} Min', value=low, key=f'{param}_min')
    high_val = st.sidebar.number_input(f'{param.capitalize()} Max', value=high, key=f'{param}_max')
    param_ranges[param] = (low_val, high_val)

# Weights for overall fitness
st.sidebar.header('Weights for Fitness Evaluation')
weights = {
    'squishiness': st.sidebar.slider('Weight for Squishiness', 0.0, 1.0, 0.3),
    'meat_taste': st.sidebar.slider('Weight for Meat Taste', 0.0, 1.0, 0.4),
    'consistency': st.sidebar.slider('Weight for Consistency', 0.0, 1.0, 0.3)
}

# Weights for meat taste evaluation
st.sidebar.header('Weights for Meat Taste Evaluation')
taste_weights = {
    'beef': st.sidebar.slider('Weight for Beef', 0.0, 1.0, 0.5),
    'garlic': st.sidebar.slider('Weight for Garlic', 0.0, 1.0, 0.3),
    'cilantro': st.sidebar.slider('Weight for Cilantro', 0.0, 1.0, 0.2)
}

# Other parameters
pop_size = st.sidebar.slider('Population Size', 50, 200, 100)
generations = st.sidebar.slider('Generations', 10, 100, 50)
mutation_rate = st.sidebar.slider('Mutation Rate', 0.01, 0.2, 0.1)
mix_factor = st.sidebar.slider('Mix Factor', 0.0, 1.0, 0.1)

if st.button('Run Simulation'):
    with st.spinner('Running genetic algorithm...'):
        best_recipe = genetic_algorithm(param_ranges, weights, pop_size, generations, mutation_rate, mix_factor, taste_weights)
    st.subheader('Best Recipe')
    for ingredient, amount in best_recipe.items():
        st.write(f'{ingredient.capitalize()}: {amount:.2f}')

    fitness = evaluate_fitness(best_recipe, weights, mix_factor, taste_weights)
    st.write(f'Fitness Score: {fitness:.2f}')