asim 25.9.7

Automatic simulation system powered by neural networks

Automatic simulation system powered by neural networks

Installation

pip install asim

What is asim

• Physical field modeling with automatic constraint enforcement
• Flexible data grouping and normalization
• Built-in support for recurrent architectures
• Export/import of trained models

Using asim

import pandas as pd
from asim.dataset import TimeSeriesDataModel, Ts, Fi, Fo
from asim.model import PhysicalFieldModel
from asim.trainer import Trainer

df = pd.read_csv("demo.csv")
cols = [
    Ts(label="ts"),
    Fi(group="boiler", label="power1", min=0.0, union="a"),
    Fi(group="boiler", label="power2", min=0.0, union="a"),
    Fo(group="boiler", label="temperature", rnn=True),
]

dm = TimeSeriesDataModel(df, columns=cols, batch_size=64)
fm = PhysicalFieldModel(dm, lr=0.003)

trainer = Trainer(max_epochs=100, enable_progress_bar=False, enable_checkpointing=False, logger=None)
trainer.fit(fm.model, datamodule=dm.datamodule)

fm.export("demo.pkl")

from asim.simulator import PhysicalSimulator
import matplotlib.pyplot as plt

simulator = PhysicalSimulator("demo.pkl", x_names=dm.x_names, y_names=dm.y_names)
x, y, y_ = simulator.predict(dm.df[100:1000], y0=[100])
plt.plot(y)
plt.plot(y_)
plt.show()