optixlog 0.3.0

Client SDK for logging photonic simulations to OptixLog with MPI support

OptixLog SDK

Experiment tracking for photonic simulations with automatic MPI support.

🚀 Quick Start

pip install optixlog

from optixlog import Optixlog

# Create client with your API key
client = Optixlog("your_api_key")

# Get or create a project
project = client.project("my_project")

# Create a run
run = project.run("experiment_1")

# Add configuration
run.add_config({"lr": 0.001, "epochs": 100})

# Log metrics
run.log(step=0, loss=0.5, accuracy=0.9)

# Log matplotlib plots
import matplotlib.pyplot as plt
fig, ax = plt.subplots()
ax.plot([1, 2, 3], [1, 4, 9])
run.log_matplotlib("my_plot", fig)

That's it! View your results at optixlog.com

📚 API Structure

from optixlog import Optixlog

# Initialize client
client = Optixlog(api_key)

# Get project (creates automatically if doesn't exist)
project = client.project("project_name_or_id")

# Create a run
run = project.run("run_name")

# Available methods on run:
run.add_config({"key": "value"})      # Add configuration
run.log(step=0, metric=value)         # Log metrics
run.log_image("key", pil_image)       # Log PIL image
run.log_file("key", "path/to/file")   # Log file
run.log_matplotlib("key", fig)         # Log matplotlib figure
run.log_plot("key", x, y)             # Create and log simple plot
run.log_array_as_image("key", array)  # Log numpy array as heatmap
run.log_histogram("key", data)        # Create and log histogram
run.log_scatter("key", x, y)          # Create and log scatter plot
run.log_batch([...])                  # Log multiple metrics in parallel

🎯 Common Use Cases

Parameter Sweep

from optixlog import Optixlog

client = Optixlog(api_key)

for lr in [0.001, 0.01, 0.1]:
    for batch_size in [16, 32, 64]:
        run = client.project("sweep").run(f"lr={lr}_bs={batch_size}")
        run.add_config({"lr": lr, "batch_size": batch_size})
        
        # Your training code here
        for step in range(100):
            loss = train_step(lr, batch_size)
            run.log(step=step, loss=loss)

Training Loop

client = Optixlog(api_key)
run = client.project("training").run("experiment_v1")

run.add_config({
    "model": "resnet50",
    "optimizer": "adam",
    "lr": 0.001
})

for epoch in range(100):
    train_loss = train_epoch()
    val_loss = validate()
    run.log(step=epoch, train_loss=train_loss, val_loss=val_loss)

Log Matplotlib Plots

import matplotlib.pyplot as plt

run = client.project("analysis").run("plots")

# Simple way
fig, ax = plt.subplots()
ax.plot(x, y)
run.log_matplotlib("my_plot", fig)

# Even simpler - create and log in one call
run.log_plot("loss_curve", steps, losses, title="Training Loss", ylabel="Loss")

Log Field Data (Numpy Arrays)

import numpy as np

run = client.project("simulation").run("field_sweep")

field = np.random.rand(100, 100)
run.log_array_as_image("field", field, cmap='hot', title="E-field intensity")

Context Manager Support

with client.project("experiment").run("run_1") as run:
    run.add_config({"param": 1})
    for step in range(100):
        run.log(step=step, loss=compute_loss())
# Automatically prints completion status

🔧 Installation

From PyPI

pip install optixlog

From Source

git clone https://github.com/yourusername/optixlog-sdk.git
cd optixlog-sdk
pip install -e .

🔑 Setup

1. Get your API key from optixlog.com
2. Use directly:

client = Optixlog("your_api_key")

3. Or set environment variable:

export OPTIX_API_KEY="your_api_key"

import os
client = Optixlog(os.getenv("OPTIX_API_KEY"))

✨ Key Features

• Fluent API: Chain methods naturally: client.project().run().log()
• Zero Boilerplate: Log matplotlib plots in one line
• Auto Project Creation: Projects created automatically if they don't exist
• Context Managers: Clean with statement support
• Input Validation: Catches NaN/Inf and invalid data
• Rich Output: Colored console feedback
• MPI Support: Automatic detection and rank 0 logging
• Batch Operations: Fast parallel uploads
• Return Values: Get URLs and status for everything

🛠️ Requirements

• Python 3.8+
• requests
• pillow
• rich (optional, for colored output)
• numpy (optional, for array logging)
• matplotlib (optional, for plot logging)

🚀 What's New in v0.2.0

• ✨ New fluent API: Optixlog → Project → Run structure
• ✨ Simplified interface: All logging methods on Run object
• ✨ Auto project creation: No need to create projects manually
• ✨ Chaining support: client.project("x").run("y").add_config({...})

📝 License

MIT License - see LICENSE file for details

---

Version: 0.2.0
Made with ⚡ for photonic simulation tracking