Skip to content

README.md

Latest commit

 

History

History
145 lines (114 loc) · 4.66 KB

README.md

File metadata and controls

145 lines (114 loc) · 4.66 KB

Questions

Ask us questions about anything related to Open Source! To add your question, create an issue in this repository.

Just a few guidelines to remember before you ask a question:

  • Ensure your question hasn't already been answered. If it has been answered but does not satisfy you, feel free to comment in the issue and we will re-open it.

  • Use a succinct title and description.

  • If your question has already been asked and answered adequately, please add a thumbs-up (or the emoji of your choice!) to the issue. This helps us in identifying common problems that people usually face.

  • Lastly, be civil and polite. :)

  • I am keep getting "Unicode Decode Error" when I run this code in jupyter lab. sometimes it works all the iterations sometimes not.... Please somebody help me....-

!pip install -U pysr import pandas as pd import os

def read_csv_to_numpy(file_path): try: # Specify the encoding explicitly to avoid UnicodeDecodeError data = pd.read_csv(file_path, header=1, encoding="UTF-8").dropna() return data except Exception as e: print(f"Failed to read {file_path} due to {e}") return pd.DataFrame() # Return empty DataFrame on failure

def load_data(file_names, data_path): file_paths = [os.path.join(data_path, name) for name in file_names] data_frames = [read_csv_to_numpy(path) for path in file_paths] return pd.concat(data_frames, ignore_index=True) if data_frames else pd.DataFrame()

Example Usage:

Update the path to your actual data directory

caljetData_path = "/Users/kevin/PSUresearch/2025PSUresearch/calijetDatacsv" data = load_data([ "0.1Ma_test1.csv", "0.1Ma_test2.csv", "0.1Ma_test3.csv", "0.1Ma_test4.csv", "0.1Ma_test5.csv", "0.1Ma_test6.csv", "0.1Ma_test7.csv", "0.1Ma_test8.csv", "0.1Ma_test9.csv", "0.1Ma_test10.csv" ], caljetData_path)

if data.empty: print("No data loaded. Exiting.") else: # Extract necessary columns X = data.iloc[:, :3].to_numpy() # Input Variables: Pitch, Yaw, Velocity pit = X[:, 0] # Input Variable: Pitch yaw = X[:, 1] # Input Variable: Yaw in_u = X[:, 2] # Input Variable: Velocity y = data.iloc[:, 9].to_numpy() # Assuming column 10 is Target Variable (Output): Pressure

print("Data loaded successfully!")
print("X shape:", X.shape)
print("y shape:", y.shape)

#Specify the data path y_truepath = '/Users/kevin/PSUresearch/2025PSUresearch/calijetDatacsv'

#Load y_true val data = load_data([ "averaged_0.1Ma.csv" ], y_truepath)

if data.empty: print("No data loaded. Exiting.") else: # Extract necessary columns Press = data.iloc[:, :8].to_numpy() # Input Variable: Pitch, Yaw, Velocity p1 = X[:, 3] p2 = X[:, 4] p3 = X[:, 5] p4 = X[:, 6] p5 = X[:, 7] pavg = X[:, 8] pstatic = X[:, 9] pstag = X[:, 10]

y = data.iloc[:, 11].to_numpy()   # Assuming column 10 is Target Variable (Output): Pressure

Example use case: Train a symbolic regression model

model = PySRRegressor( model_selection='best', unary_operators=["cos", "sin", "square", "inv(x) = 1/x", "exp"], binary_operators=["+", "-", "/", "*"], extra_sympy_mappings={"inv": lambda x: 1/x}, niterations=400, populations=30 ) X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2) model.fit(X_train, y_train) #model.get_best().equation print(model.equations_) def elementwise_loss(y_true, y_pred):

# Convert inputs to numpy arrays
y_true = np.array(y_true)
y_pred = np.array(y_pred)

# Element-wise squared error
squared_error = (y_true - y_pred) ** 2

# Element-wise absolute error
absolute_error = np.abs(y_true - y_pred)

# MSE: mean of squared_error over axis=0
mse = np.mean(squared_error, axis=0)

# RMSE: square root of MSE
rmse = np.sqrt(mse)

# MAE: mean of absolute_error
mae = np.mean(absolute_error, axis=0)

# Scatter loss: Logarithmic scaled error
# (use small epsilon for numerical stability)
epsilon = 1e-20
scatter_loss = np.abs(np.log((np.abs(y_pred) + epsilon) / (np.abs(y_true) + epsilon)))

# Sign loss: Penalize differences in signs
sign_loss = 10 * (np.sign(y_pred) - np.sign(y_true)) ** 2

# Combined loss: sum of scatter_loss and sign_loss
combined_loss = scatter_loss + sign_loss

# Return a dictionary of metrics
return {
    "MSE": mse,
    "RMSE": rmse,
    "MAE": mae,
    "Scatter Loss": np.mean(scatter_loss),
    "Sign Loss": np.mean(sign_loss),
    "Combined Loss": np.mean(combined_loss)
}

y_pred =

#Call the function losses = elementwise_loss(y_true, y_pred)

#Print out the results print("Element-wise Losses:") for key, value in losses.items():