arXiv subjects

Michael Petrov

Publications and source records attributed to Michael Petrov.

13 recordsLinked to original sources

arXiv2026

Irradiation-Induced Spin Bath Evolution and as-Grown Hydrogen Defects in CVD Diamond Revealed by NV-Based DEER Spectroscopy

The aim of this paper is to provide the reader with a review and current state of the art of the fabrication of high T2 coherence diamond, optimised by the use of double electron-electron resonance (DEER) spectroscopy. Using DEER, we study the formation, transformation, and annealing of paramagnetic defects in as-grown CVD diamond and after post-processing. Electron irradiation leads to the formation of an additional S = 1/2 resonance in the DEER spectrum, which we consider to be a composite X ensemble. By tracking the concentrations of X and P1 point defects during annealing from 650C to 1200C, we find that the X ensemble initially consists of a mixture of V- spins and interstitial spins, which disappear at about 650C. Vacancies migrate during annealing, forming clusters that persist to 1000C and disappear upon annealing at 1200C, contributing to the X ensemble signal. We have developed a model of the influence of the mixed spin bath on the coherence of NV centers, which includes independent couplings with P1 centers, V-, divacancies, and interstitials. Detailed DEER studies allowed us to reveal and resolve a weak signal from two additional S = 1/2 species associated with hydrogen: NVH- and consistent with a substitutional hydrogen defect, which overlaps the vacancy spectral line. Taken together, these results show that the NV-DEER method is a powerful tool for investigating paramagnetic defects in diamond with high precision and nanoscale resolution, essential for material optimisation. The achieved high T2 coherence time is consistent with the spin bath model, and the crystals reach the quality required for advanced quantum sensing applications.

quant-ph↗

arXiv2026

High fidelity quantum state tomography of electron-$^{14}$N nuclear hybrid spin register in diamond using Rabi oscillations

We report on a new quantum state characterisation method, which we call Rabi-based Quantum State Tomography (RQST), that we have validated on single-qubit quantum states, in particular on the electron and nuclear spins of a single nitrogen-vacancy (NV) centre in diamond, demonstrating high fidelities. The difference of RQST with conventional tomography methods is in the implementation of rotation operators and construction of density matrix from the measured data sets. We demonstrate efficient quantum state control of the electron spin at room temperature with an average fidelity of 0.995 over more than 40 measurements on different states on the Bloch sphere with a maximum fidelity of 0.99992. Also, we apply the methodology to the dark NV nuclear spin state. The state is read via the electron spin using the C-NOT two-qubit entanglement gate and demonstrate fidelities of the same order.

quant-ph↗

arXiv2026

Long Spin Relaxation Times in CVD-Grown Nanodiamonds

Currently, the primary applications of fluorescent nanodiamonds (FNDs) are in the area of biosensing, by using photoluminescence or spin properties of colour centres, mainly represented by the Nitrogen Vacancy (NV) point defect. The sensitivity of NV-FNDs to external fields is, however, limited by crystallographic defects, which influence their key quantum state characteristics - the spin longitudinal (\textit{T$_1$}) and spin transversal (\textit{T$_2$}) relaxation and coherence times, respectively. We report on utilising an advanced FND growth technique consisting of heterogeneous nucleation on pre-engineered sites to create FNDs averaging around 60 nm in size, with mean longitudinal coherence times of 800 $μ$s and a maximum over 1.8 ms, close to bulk theoretical values. This is a major, nearly ten-fold improvement over commercially available nanodiamonds for the same size range of 50 to 150 nm. Heavy-N doped nanodiamond shells, important for sensing events in nm proximity to the diamond surface, are fabricated and discussed in terms of re-nucleation and twinning on \{111\} crystal facets. We also discuss scalability issues in order to enable the production of FND volumes matching the needs of sensing applications.

cond-mat.mes-hall↗

arXiv2026

High PDMR contrast in single NV centres and related photocurrent properties

This paper aims to extend the understanding of the mechanism of photo-electrical detection of magnetic resonance (PDMR) in nitrogen-vacancy (NV) centres. This technique is particularly important for development of solid-state quantum computing platforms. In particular, we report on the new insight in the photocurrent (PC) generation and charge cycling in the single NV centre, which is related to PDMR contrast reaching 50\% and above. We develop a technique to locate PC related features. We find that electrons generated at the NV centre are stored in interface trap levels and establish that the interface states serve as an amplifier that can be driven by introducing a second laser into our confocal setup. We show that controlling these interface states allows one to significantly enhance the PDMR contrast. We develop a model that consistently explains observed amplification effects even without the application of a bias voltage.

quant-ph↗

arXiv2026

High Fidelity Single-NV Qubit Quantum State Tomography by Photoelectric Readout

Quantum computing is a rapidly developing field. However, the most commonly used qubits require cryogenic conditions to operate, which increases the costs and puts constraints on the up-scaling. Ambient solid-state qubits provide an alternative with potential for large-scale application. The nitrogen-vacancy (NV) center in diamond is one of the main candidates for solid-state computing architectures at room temperature and has proven to be competitive in terms of gate fidelity, quantum error correction, couplings, etc. Each NV center has an associated electronic spin that is conventionally read out by photoluminescence. However, regarding the creation of small, ambient NV-based quantum processors, the optical readout introduces limitations on the collection efficiency and resolution of the readout as well as the size of the final device and its integration into standard semiconductor architectures. In this work, we investigate the competitiveness of the photoelectric readout versus the traditional optical readout. In particular, we report on using photoelectrical detection to perform quantum state tomography measurements on a single NV center. We achieve the fidelity $0.995 \pm 0.0062$ for state reconstruction, comparable to optical measurements, demonstrating that the fidelity does not suffer from the adapted readout, highlighting the value of photoelectric detection for NV-based quantum processors.

quant-ph↗

arXiv2026

Modelling and experimental verification of photoelectrical response of NV diamond spin centres

We report on a mathematical model of the photoelectric response of NV colour centres in diamond, that can be employed for sensing and quantum science information applications. Although the model applies to NV centre in diamond, it can be applied with small modifications to other semiconducting solid state qubits. In our model, we include the drift and collection of charge carriers as well as the presence of other defects via generation and recombination dynamics. Though the photoluminescence readout and the associated dynamics of the NV defect has been extensively studied experimentally and theoretically, so far, there has been no precise model for photocurrent readout, including these effects. In our description, we use a multilevel-level system including mS=0, mS=+-1 ground and excited states, singlet state and the NV0 neutral state. Also, the presence of substitutional nitrogen (NS), which for example determines the spin coherence via the paramagnetic spin bath, is discussed together with presence of acceptor defects. We model the time-dependent occupation of all electronic sublevels and also consider the electronic charge transport from the Boltzmann transport equation, leading to information about the charge state transitions and recombination dynamics. ODMR and PDMR response as well as their quantum efficiencies, are calculated. On this basis, we determine an optimal parameter space for qubit operations, including the highest spin contrast and especially relate those to NS presence. The model is confirmed experimentally and can become a useful tool for optimisation of the performance of NV qubit photoelectric readout.

quant-ph↗

arXiv2025

Charge-state stability of single NV centers in HPHT-type IIa diamond

This is a preliminary version. Improvements and additional analysis will be included in a revised manuscript. We investigate the charge-state stability of individual nitrogen-vacancy (NV) centers in weakly doped HPHT IIa diamond containing sub-ppm concentrations of boron and nitrogen. Using Ti/Al coplanar electrodes on an oxygen-terminated surface, we study how applied electric fields and optical excitation jointly govern NV charge conversion. By combining voltage-dependent photoluminescence, real-time charge-state monitoring, laser-power saturation with spectral decomposition, and time-resolved measurements, we reveal that electric fields several micrometers from the contacts significantly increase the NV- population and enhance spin readout. At low excitation powers, the NV- population evolves on minute timescales following compressed-exponential kinetics, consistent with slow space-charge rearrangement in ultra-insulating diamond. Under pulsed excitation, we observe hundreds-of-nanoseconds NV-/NV0 conversion driven by hole capture, which is strongly suppressed by applied bias. Our results demonstrate that residual boron acceptors play a key role in determining charge-state stability and show how electrical bias can reliably stabilize NV- in weakly doped bulk diamond.

quant-ph↗

arXiv2025

Electrical Readout of Spin Environments in Diamond for Quantum Sensing

Nitrogen-vacancy (NV) centres in diamond are a key platform for quantum sensing and quantum information, combining long coherence times with controllable spin-spin interactions. Most of current quantum algorithms rely on optical access, which limit device integration and applicability in opaque or miniaturized settings. Here we demonstrate an all-electrical approach, photocurrent double electron-electron resonance (PC-DEER), permitting exploiting local dipolar interactions between individual NV spin qubits or ensembles and nearby paramagnetic defects with sub-confocal resolution. PC-DEER extends photocurrent NV readout from single-spin to spin-bath control and coherent manipulation, enabling characterization of bath-induced noise and effective deployment of noise-reduction protocols. We resolve the signatures of substitutional nitrogen (P1) and NVH centers with reproducible contrast by using electrical signals. Our results establish a scalable, optical-free spin readout strategy that bridges fundamental studies of spin environments with deployable quantum technologies, advancing the integration of diamond-based sensors into solid-state quantum devices.

quant-ph↗

arXiv2025

Power Stabilization for AI Training Datacenters

Large Artificial Intelligence (AI) training workloads spanning several tens of thousands of GPUs present unique power management challenges. These arise due to the high variability in power consumption during the training. Given the synchronous nature of these jobs, during every iteration there is a computation-heavy phase, where each GPU works on the local data, and a communication-heavy phase where all the GPUs synchronize on the data. Because compute-heavy phases require much more power than communication phases, large power swings occur. The amplitude of these power swings is ever increasing with the increase in the size of training jobs. An even bigger challenge arises from the frequency spectrum of these power swings which, if harmonized with critical frequencies of utilities, can cause physical damage to the power grid infrastructure. Therefore, to continue scaling AI training workloads safely, we need to stabilize the power of such workloads. This paper introduces the challenge with production data and explores innovative solutions across the stack: software, GPU hardware, and datacenter infrastructure. We present the pros and cons of each of these approaches and finally present a multi-pronged approach to solving the challenge. The proposed solutions are rigorously tested using a combination of real hardware and Microsoft's in-house cloud power simulator, providing critical insights into the efficacy of these interventions under real-world conditions.

cs.AR↗

arXiv2024

High fidelity two-qubit quantum state tomography of Electron-14N hybrid spin register in diamond

We report here on a major improvement of the control and characterization capabilities of 14N nuclear spin of single NV centers in diamond, as well as on a new method that we have devised for characterizing quantum states, i.e. quantum state tomography using Rabi experiments. Depending on whether we use amplitude information or phase information from Rabi experiments, we define two sub-methods namely Rabi amplitude quantum state tomography (RAQST) and Rabi phase quantum state tomography (RPQST). The advantage of Rabi-based tomography methods is that they lift the requirement of unitary operations used in other methods in general and standard methods in particular. On one hand, this does not increase the complexity of the tomography experiments in large registers, and on the other hand, it decreases the error induced by MW irradiation. We used RAQST and RPQST to investigate the quality of various two-qubit pure states in our setup. As expected, test quantum states show very high fidelity with the theoretical counterpart.

quant-ph↗

arXiv2021

Evaluating Large Language Models Trained on Code

We introduce Codex, a GPT language model fine-tuned on publicly available code from GitHub, and study its Python code-writing capabilities. A distinct production version of Codex powers GitHub Copilot. On HumanEval, a new evaluation set we release to measure functional correctness for synthesizing programs from docstrings, our model solves 28.8% of the problems, while GPT-3 solves 0% and GPT-J solves 11.4%. Furthermore, we find that repeated sampling from the model is a surprisingly effective strategy for producing working solutions to difficult prompts. Using this method, we solve 70.2% of our problems with 100 samples per problem. Careful investigation of our model reveals its limitations, including difficulty with docstrings describing long chains of operations and with binding operations to variables. Finally, we discuss the potential broader impacts of deploying powerful code generation technologies, covering safety, security, and economics.

cs.LG↗

arXiv2019

Dota 2 with Large Scale Deep Reinforcement Learning

On April 13th, 2019, OpenAI Five became the first AI system to defeat the world champions at an esports game. The game of Dota 2 presents novel challenges for AI systems such as long time horizons, imperfect information, and complex, continuous state-action spaces, all challenges which will become increasingly central to more capable AI systems. OpenAI Five leveraged existing reinforcement learning techniques, scaled to learn from batches of approximately 2 million frames every 2 seconds. We developed a distributed training system and tools for continual training which allowed us to train OpenAI Five for 10 months. By defeating the Dota 2 world champion (Team OG), OpenAI Five demonstrates that self-play reinforcement learning can achieve superhuman performance on a difficult task.

cs.LG↗

arXiv2010

Classification of $k$-tangle projections using cascade representation

The paper addresses the $k$-tangle enumeration problem. We introduce a notion of cascade diagram for $k$-tangle projections. An effective enumeration algorithm for projections is proposed based on cascade representation. Tangles projections with up to 12 crossings are tabulated. We provide also pictures of alternating $k$-tangles with 5 crossing or less.

math.GT↗