Bayesian NMOR (Nonlinear Magneto Optical Magnetometry)

A research project implementing Bayesian inference approaches for Nonlinear Magneto Optical Magnetometry using adaptive experimental design and real-time hardware control.

Overview

This project combines theoretical Bayesian inference methods with practical hardware implementation to optimize magnetic field measurements using nonlinear magneto-optical rotation (NMOR). The system uses adaptive experimental design to iteratively select optimal bias fields that maximize information gain about unknown magnetic fields.

Project Structure

bayesian_nmor/
├── BayesianNMOR/              # Core Bayesian inference implementation
│   ├── bayes_experiment.py    # Main experimental utilities and algorithms
│   ├── toy_bayesian_model.ipynb  # Interactive notebook with examples
│   └── tests/                 # Unit tests and test documentation
├── FPGA/                      # Hardware control and data acquisition
│   ├── fpga.py               # Red Pitaya FPGA control via SCPI
│   └── scpi_loadtest.py      # SCPI communication performance testing
├── requirements.txt          # Python dependencies
└── README.md                # This file

Key Features

Bayesian Inference Engine (BayesianNMOR/)

• Adaptive Experimental Design: Automatically selects optimal bias fields to maximize information gain
• Dispersive Lorentzian Signal Model: Implements the theoretical model for NMOR signals
• Real-time Posterior Updates: Bayesian updating with Gaussian measurement noise
• Information-theoretic Optimization: Uses KL divergence to quantify and maximize information gain

Hardware Integration (FPGA/)

• Real-time Control: Direct communication with Red Pitaya FPGA boards via SCPI protocol
• PID Controller Interface: Configure and control feedback loops for magnetic field stabilization
• Data Acquisition: Oscilloscope functionality for capturing measurement traces
• Performance Monitoring: Latency testing and system optimization tools

Quick Start

Installation

1. Clone the repository:

git clone https://github.com/apranav22/bayesian_nmor.git
cd bayesian_nmor

2. Install dependencies:

pip install -r requirements.txt

3. Run tests to verify installation:

pytest -q

Basic Usage

Running Bayesian Experiments

from BayesianNMOR.bayes_experiment import run_experiment

# Run an adaptive Bayesian experiment
B_grid, posteriors, biases = run_experiment(
    N_exp=5,           # Number of experimental iterations
    B_true=0.2,        # True unknown field (for simulation)
    Gamma=1.0,         # Lorentzian width parameter
    sigma=0.05,        # Measurement noise standard deviation
    B_range=(-1.0, 1.0), # Field search range
    n_grid=500         # Grid resolution for posterior
)

print("Optimal bias fields chosen:", biases)

Hardware Control

from FPGA.fpga import set_pid, set_setpoint, capture_trace

# Configure PID controller
set_pid(kp=0.3, ki=1.0, kd=0.0)

# Set field setpoint
set_setpoint(0.2)  # 0.2 V setpoint

# Capture measurement trace
capture_trace("measurement.csv", dec=64)

Interactive Exploration

Open the Jupyter notebook for interactive examples:

jupyter notebook BayesianNMOR/toy_bayesian_model.ipynb

Core Algorithms

Signal Model

The system models NMOR signals using a dispersive Lorentzian:

α(B) = B_total / (B_total² + Γ²)

where B_total = B_unknown + B_bias and Γ is the resonance width.

Adaptive Design

The algorithm selects bias fields B_bias that maximize expected information gain:

1. Prior Update: Start with uniform prior over field range
2. Utility Calculation: Compute expected KL divergence for candidate bias fields
3. Optimal Selection: Choose bias that maximizes expected information gain
4. Measurement: Acquire noisy measurements at selected bias
5. Posterior Update: Apply Bayes' rule to update field estimate
6. Iteration: Repeat until convergence or iteration limit

Hardware Requirements

• Red Pitaya FPGA Board: For real-time control and data acquisition
• Network Connection: Ethernet connection to FPGA board
• Magnetic Field Generation: Coils or other field sources controlled via PID

Testing

The project includes comprehensive unit tests covering:

• Signal model properties (odd function, boundedness)
• Likelihood function correctness
• Posterior normalization and convergence
• Information-theoretic measures
• End-to-end experimental workflows

Run the full test suite:

pytest tests/ -v

For testing documentation, see BayesianNMOR/tests/TESTING.md.

License

[Add appropriate license information]

Citation

Contact

Author: Dinesh R, Pranav Agrawal Repository: https://github.com/apranav22/bayesian_nmor

For questions or collaboration inquiries, please open an issue on GitHub.