Efficient adaptive Bayesian estimation of a slowly fluctuating Overhauser field gradient

Jacob Benestad,Jan A. Krzywda,Evert van Nieuwenburg,Jeroen Danon

2024-06-12

Abstract:Slow fluctuations of Overhauser fields are an important source for decoherence in spin qubits hosted in III-V semiconductor quantum dots. Focusing on the effect of the field gradient on double-dot singlet-triplet qubits, we present two adaptive Bayesian schemes to estimate the magnitude of the gradient by a series of free induction decay experiments. We concentrate on reducing the computational overhead, with a real-time implementation of the schemes in mind. We show how it is possible to achieve a significant improvement of estimation accuracy compared to more traditional estimation methods. We include an analysis of the effects of dephasing and the drift of the gradient itself.

Mesoscale and Nanoscale Physics,Quantum Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **How to efficiently estimate the Overhauser field gradient of the double - dot singlet - triplet spin qubit in III - V semiconductor quantum dots in real - time?** Specifically, the author focuses on the spin - dephasing problem caused by the slow fluctuation of the Overhauser field gradient, and proposes two adaptive Bayesian estimation schemes to improve the estimation accuracy and reduce the computational cost.

### Problem Background

In III - V semiconductor quantum dots, spin qubits (such as GaAs) are affected by the nuclear spin bath, resulting in a randomly fluctuating Overhauser field. These fluctuations will limit the spin - dephasing time ($T_2^*$), which is usually around 20 nanoseconds. In order to prolong the coherence time of spin qubits, it is crucial to accurately estimate and monitor the Overhauser field gradient in real - time. Moreover, if real - time estimation can be achieved, the Overhauser field can also be used as a control axis, thereby improving the operation accuracy of qubits.

### Research Objectives

1. **Improve Estimation Accuracy**: Significantly improve the estimation accuracy of the Overhauser field gradient through adaptive Bayesian estimation compared with traditional estimation methods.
2. **Reduce Computational Cost**: Design an adaptive Bayesian estimation scheme suitable for real - time implementation, considering the computational complexity and information storage capacity limitations of current experimental equipment.
3. **Handle Slowly Drifting Parameters**: Analyze the influence of the dephasing effect and the gradient drift itself on the estimation, and ensure the effectiveness of the estimation scheme.

### Method Overview

The author proposes two adaptive Bayesian estimation schemes:

1. **Moment - method Fitting**:
- Use the second - order and fourth - order moments to fit the posterior distribution as a bimodal Gaussian distribution.
- Calculate the expected variance through an analytical expression and select the optimal waiting time $\tau_n$.

2. **KL - Divergence Fitting Based on Neural Network**:
- Use a neural network to minimize the KL - divergence between the true posterior distribution and the bimodal Gaussian distribution.
- Generate a training data set and train the neural network to predict the updated parameters $\mu_n, \sigma_n$ and the optimal waiting time $\tau_n$.

### Results and Conclusions

By performing numerical simulations on these two schemes and comparing them with non - adaptive methods, the author demonstrates the superior performance of the adaptive Bayesian estimation schemes in estimating static and slowly drifting parameters. In particular, when dealing with slowly changing Overhauser field gradients, these schemes can effectively reduce the estimation error and provide more accurate predictions.

### Formula Summary

1. **Second - Order Moment Formula**:
\[
\langle \omega^2 \rangle_n = \frac{\Re\left[f_{n - 1}(0)+(- 1)^{d_n}f_{n - 1}(\tau)\right]}{\Re\left[\nu_{n - 1}(0)+(- 1)^{d_n}\nu_{n - 1}(\tau)\right]}
\]
\[
\langle \omega^4 \rangle_n = \frac{\Re\left[g_{n - 1}(0)+(- 1)^{d_n}g_{n - 1}(\tau)\right]}{\Re\left[\nu_{n - 1}(0)+(- 1)^{d_n}\nu_{n - 1}(\tau)\right]}
\]

2. **Expected Variance Formula**:
\[
E_{d_n}\left[\hat{\sigma}_n^2\right]=\frac{1}{2}\left[\left(\hat{\sigma}_n^{(0)}\right)^2-\left(\hat{\sigma}_n^{(1)}\right)^2\right]e^{-\frac{1}{2}\alpha_{n - 1}^2\tau^2}\cos(\mu_{n - 1}\tau)+\left(\hat{\sigma}_n^{(0)}\right)^2+\left(\hat{\sigma}_n^{(1)}\right)^2
\]