KoopmanESN/TrajTrans/StateOnly/ESN/KoopmanESNTest.py

import torch
from KoopmanESNModel import KoopmanESNConfig, KoopmanESN
from BaseModel.ESNModel import ESN
from DataProcess import DataLoad
import numpy as np
import matplotlib.pyplot as plt
import time

KoopmanModel = torch.load('./ModelLib/DeepKoopmanModel.pt', map_location="cpu")
encoder = KoopmanModel.encoder
decoder = KoopmanModel.decoder
A = KoopmanModel.A.detach().numpy()
state_dim = KoopmanModel.state_dim
latent_dim = KoopmanModel.latent_dim

MonteNum = 100
MinStart = 0
MaxStart = 20000
StartList = np.random.randint(MinStart, MaxStart, size=MonteNum)
ESNMSEList = []
KoopMSEList = []
KoopTime = 0
ESNTime = 0
count = 0

lr = 0.5
sr = 0.9
sp = 0.1
ridge = 1e-7
# 短时预测
# train_len = 500
# train_warmup = 10
# predict_warmup = 100
# predict_len = 20
# 长期预测
train_len = 5000
train_warmup = 100
predict_warmup = 100
predict_len = 1000

for train_start in StartList:
    count += 1
    print(count)

    Koopmanconfig = KoopmanESNConfig(
        units=latent_dim,
        lr=lr,
        sr=sr,
        sp=sp,
        ridge=ridge,
        train_start=train_start,
        train_len=train_len,
        train_warmup=train_warmup,
        predict_warmup=predict_warmup,
        predict_len=predict_len,
        state_dim=latent_dim
    )

    ESNconfig = KoopmanESNConfig(
        units=1000,
        lr=lr,
        sr=sr,
        sp=sp,
        ridge=ridge,
        train_start=train_start,
        train_len=train_len,
        train_warmup=train_warmup,
        predict_warmup=predict_warmup,
        predict_len=predict_len,
        state_dim=latent_dim
    )

    KoopESNModel = KoopmanESN(
        config=Koopmanconfig,
        A=A
    )
    ESNModel = ESN(
        config=ESNconfig
    )

    _, DataTrain, DataWarm, DataVal = DataLoad('./DataLib/LatentNor.csv', Koopmanconfig)
    # _, DataTrain, DataWarm = DataLoad('./DataLib/DataNor.csv', config)

    # KoopmanESN
    time_start = time.time()
    KoopESNModel.koopmanesn_train(DataTrain)
    DataPreKoop = KoopESNModel.predict(DataWarm)
    time_end = time.time()
    KoopTime += time_end-time_start

    # ESN
    time_start = time.time()
    ESNModel.esn_train(DataTrain)
    DataPreESN = ESNModel.predict(DataWarm)
    time_end = time.time()
    ESNTime += time_end - time_start

    DataPreKoop_Ten = torch.from_numpy(DataPreKoop).float()
    DataPreESN_Ten = torch.from_numpy(DataPreESN).float()
    DataVal_Ten = torch.from_numpy(DataVal).float()
    DataWarm_Ten = torch.from_numpy(DataWarm).float()

    StatePreKoop_Ten = decoder(DataPreKoop_Ten)
    StatePreESN_Ten = decoder(DataPreESN_Ten)
    StateVal_Ten = decoder(DataVal_Ten)
    StateWarm_Ten = decoder(DataWarm_Ten)

    StatePreKoop = StatePreKoop_Ten.detach().numpy()
    StatePreESN = StatePreESN_Ten.detach().numpy()
    StateVal = StateVal_Ten.detach().numpy()
    StateWarm = StateWarm_Ten.detach().numpy()

    KoopMSE = np.linalg.norm(StatePreKoop - StateVal, ord='fro') ** 2 / np.prod(StateVal.shape)
    ESNMSE = np.linalg.norm(StatePreESN - StateVal, ord='fro') ** 2 / np.prod(StateVal.shape)

    KoopMSEList.append(KoopMSE)
    ESNMSEList.append(ESNMSE)

KoopTime /= MonteNum
ESNTime /= MonteNum

KoopMSEList = np.log10(KoopMSEList)
ESNMSEList = np.log10(ESNMSEList)

print(f"KoopESN平均单次训练+预测耗时:{KoopTime:.4f}s")
print(f"ESN平均单次训练+预测耗时:{ESNTime:.4f}s")

plt.figure(figsize=(8, 5))

# 绘制箱线图（先使用统一的boxprops）
box = plt.boxplot(
    [KoopMSEList, ESNMSEList],
    labels=['KoopmanESN', 'ESN'],
    patch_artist=True,
    boxprops={'facecolor': 'lightblue', 'edgecolor': 'navy'},  # 初始统一设置
    medianprops={'color': 'red', 'linewidth': 2},
    whiskerprops={'color': 'gray', 'linestyle': '--'},
    flierprops={'marker': 'o', 'markersize': 5}
)

# 绘制后单独设置每个箱体的颜色
colors = ['lightblue', 'lightgreen']
edge_colors = ['navy', 'darkgreen']

for box_element, color, edge_color in zip(box['boxes'], colors, edge_colors):
    box_element.set_facecolor(color)
    box_element.set_edgecolor(edge_color)

# 添加图表元素
plt.title('Model Training Loss Comparison (Boxplot)', fontsize=12)
plt.ylabel('log10 Loss Value', fontsize=10)
plt.legend([box["boxes"][0], box["boxes"][1]], ['KoopmanESN', 'ESN'])
plt.grid(True, linestyle='--', alpha=0.3)

plt.tight_layout()
plt.show()