arXiv subjects

Harsh Mishra

Publications and source records attributed to Harsh Mishra.

11 recordsLinked to original sources

arXiv2026

Interface-engineered oxidation-resistant wafer-level Tantalum-Tantalum thermocompression bonding for 3D integration of superconducting interconnects

Wafer-level 3D integration of superconducting interconnects requires low-thermal budget bonding processes compatible with superconducting materials. Rapid native oxide formation on tantalum (Ta) surfaces limits low-temperature, low-pressure direct Ta-Ta thermocompression bonding. In this study, we develop an oxidation-resistant bonding process using an ultrathin Au passivation layer to suppress oxide formation during bonding. The engineered interface enables blanket Ta-Ta wafer bonding at 300 $^\circ$C under 4.93 bar, significantly reducing the bonding thermal budget and generation of $\alpha$-Ta across the interface, which potentially improves coherence time as reported in literature. Structural and interfacial analyses confirm oxide suppression and continuous metallic bonding, having a bond strength of 169 MPa. This work demonstrates a low-temperature, low-pressure Ta-Ta thermocompression bonding strategy for scalable 3D superconducting interconnect integration.

cond-mat.supr-con↗

arXiv2026

Efficient $n$-qubit entangling operations via a superconducting quantum router

Quantum algorithms on near-term quantum processors are typically executed using shallow quantum circuits composed of one- and two-qubit gates. However, as circuit depth and gate number increase, gate imperfections and qubit decoherence begin to dominate, limiting algorithmic complexity. An alternative approach is to explore gates involving more than two qubits. In previous work (X. Wu et al., Physical Review X 14, 041030 (2024)), we demonstrated a new superconducting qubit architecture with user-selectable two-qubit interactions via a reconfigurable router, used to connect pairs of qubits. Here, we leverage this novel architecture to realize programmable and efficient multi-qubit operations involving more than two qubits, resulting in faster preparation of multi-qubit entangled states with good fidelities. We also successfully apply model-free reinforcement learning to perform multi-qubit gates, including training a two-qubit controlled-Z gate as well as three-qubit controlled-SWAP and controlled-controlled-phase (Fredkin and Toffoli) gates. Higher $n$th-order gates may also be feasible, using our high-connectivity router design. This could provide a more efficient and higher-fidelity implementation of complex quantum algorithms and a more practical approach to quantum computation.

quant-ph↗

arXiv2026

Large Language Model-Assisted Superconducting Qubit Experiments

Superconducting circuits have demonstrated significant potential in quantum information processing and quantum sensing. Implementing novel control and measurement sequences for superconducting qubits is often a complex and time-consuming process, requiring extensive expertise in both the underlying physics and the specific hardware and software. In this work, we introduce a framework that leverages a large language model (LLM) to automate qubit control and measurement. Specifically, our framework conducts experiments by generating and invoking schema-less tools on demand via a knowledge base on instrumental usage and experimental procedures. We showcase this framework with two experiments: an autonomous resonator characterization and a direct reproduction of a quantum non-demolition (QND) characterization of a superconducting qubit from literature. This framework enables rapid deployment of standard control-and-measurement protocols and facilitates implementation of novel experimental procedures, offering a more flexible and user-friendly paradigm for controlling complex quantum hardware.

quant-ph↗

arXiv2025

Demonstration of surface-engineered oxidation-resistant Nb-Nb thermocompression bonding toward scalable superconducting quantum computing architectures

Scalable quantum computing currently requires a large array of qubit integration, but present two-dimensional interconnects face challenges such as wiring congestion, electromagnetic interference, and limited cryogenic space. To overcome this challenge, implementing three-dimensional (3D) vertical architectures becomes crucial. Niobium (Nb), due to its excellent superconducting characteristics and strong fabrication process compatibility, stands out as a prime material choice. The main challenge in Nb-Nb bonding is the presence of an oxide layer at the interface, even after post-bonding annealing across various bonding methods. The native Nb oxide forms rapidly in air, creating a resistive barrier to supercurrent flow and introducing two-level system losses that degrade qubit coherence while increasing the overall thermal budget. These issues show the need for effective surface engineering to suppress oxidation during bonding. This study introduces an ultrathin gold (Au) capping layer as a passivation strategy to prevent oxygen incorporation at the Nb surface. This approach enables low-temperature Nb-Nb thermocompression bonding at 350 {\deg}C under a reduced bonding pressure of 0.495 MPa. Detailed microstructural and interfacial analyses confirm that Au passivation effectively suppresses oxide formation and hence enhances bonding uniformity and strength with keeping the superconductivity, establishing a robust route toward low-temperature, low-pressure Nb-Nb bonding for scalable 3D superconducting quantum computing architectures.

cond-mat.supr-con↗

arXiv2025

Ghost Imaging with Free Electron-Photon Pairs

Coincidence imaging, also known as ghost imaging, is a technique that exploits correlations between two particles to reconstruct information about a specimen. The particle that relays the spatial information about the object remains completely non-interacting, while the particle used to probe the object is not spatially resolved. While ghost imaging has been primarily implemented on photonic platforms, it becomes particularly intriguing when applied to particles with fundamentally different properties, such as massive, charged electrons and massless, neutral photons, especially considering the role of both particles as cornerstones of highly advanced microscopic platforms. In this work, we investigate coincidence imaging using electron-cathodoluminescence photon pairs generated within a transmission electron microscope. Utilizing a custom-built free-space cathodoluminescence setup, we demonstrate ghost imaging of complex patterns. We are able to obtain a spatial resolution down to 2 $\mu$m, paving the way for adaptation of quantum-enhanced imaging techniques from photonic quantum optics to electron microscopy.

quant-ph↗

arXiv2025

Mitigating cosmic ray-like correlated events with a modular quantum processor

Quantum processors based on superconducting qubits are being scaled to larger qubit numbers, enabling the implementation of small-scale quantum error correction codes. However, catastrophic chip-scale correlated errors have been observed in these processors, attributed to e.g. cosmic ray impacts, which challenge conventional error-correction codes such as the surface code. These events are characterized by a temporary but pronounced suppression of the qubit energy relaxation times. Here, we explore the potential for modular quantum computing architectures to mitigate such correlated energy decay events. We measure cosmic ray-like events in a quantum processor comprising a motherboard and two flip-chip bonded daughterboard modules, each module containing two superconducting qubits. We monitor the appearance of correlated qubit decay events within a single module and across the physically separated modules. We find that while decay events within one module are strongly correlated (over $85\%$), events in separate modules only display $\sim 2\%$ correlations. We also report coincident decay events in the motherboard and in either of the two daughterboard modules, providing further insight into the nature of these decay events. These results suggest that modular architectures, combined with bespoke error correction codes, offer a promising approach for protecting future quantum processors from chip-scale correlated errors.

quant-ph↗

arXiv2023

Accelerated Neural Network Training with Rooted Logistic Objectives

Many neural networks deployed in the real world scenarios are trained using cross entropy based loss functions. From the optimization perspective, it is known that the behavior of first order methods such as gradient descent crucially depend on the separability of datasets. In fact, even in the most simplest case of binary classification, the rate of convergence depends on two factors: (1) condition number of data matrix, and (2) separability of the dataset. With no further pre-processing techniques such as over-parametrization, data augmentation etc., separability is an intrinsic quantity of the data distribution under consideration. We focus on the landscape design of the logistic function and derive a novel sequence of {\em strictly} convex functions that are at least as strict as logistic loss. The minimizers of these functions coincide with those of the minimum norm solution wherever possible. The strict convexity of the derived function can be extended to finetune state-of-the-art models and applications. In empirical experimental analysis, we apply our proposed rooted logistic objective to multiple deep models, e.g., fully-connected neural networks and transformers, on various of classification benchmarks. Our results illustrate that training with rooted loss function is converged faster and gains performance improvements. Furthermore, we illustrate applications of our novel rooted loss function in generative modeling based downstream applications, such as finetuning StyleGAN model with the rooted loss. The code implementing our losses and models can be found here for open source software development purposes: https://anonymous.4open.science/r/rooted_loss.

cs.LG↗

arXiv2023

Flag Aggregator: Scalable Distributed Training under Failures and Augmented Losses using Convex Optimization

Modern ML applications increasingly rely on complex deep learning models and large datasets. There has been an exponential growth in the amount of computation needed to train the largest models. Therefore, to scale computation and data, these models are inevitably trained in a distributed manner in clusters of nodes, and their updates are aggregated before being applied to the model. However, a distributed setup is prone to Byzantine failures of individual nodes, components, and software. With data augmentation added to these settings, there is a critical need for robust and efficient aggregation systems. We define the quality of workers as reconstruction ratios $\in (0,1]$, and formulate aggregation as a Maximum Likelihood Estimation procedure using Beta densities. We show that the Regularized form of log-likelihood wrt subspace can be approximately solved using iterative least squares solver, and provide convergence guarantees using recent Convex Optimization landscape results. Our empirical findings demonstrate that our approach significantly enhances the robustness of state-of-the-art Byzantine resilient aggregators. We evaluate our method in a distributed setup with a parameter server, and show simultaneous improvements in communication efficiency and accuracy across various tasks. The code is publicly available at https://github.com/hamidralmasi/FlagAggregator

cs.LG↗

arXiv2023

Using Intermediate Forward Iterates for Intermediate Generator Optimization

Score-based models have recently been introduced as a richer framework to model distributions in high dimensions and are generally more suitable for generative tasks. In score-based models, a generative task is formulated using a parametric model (such as a neural network) to directly learn the gradient of such high dimensional distributions, instead of the density functions themselves, as is done traditionally. From the mathematical point of view, such gradient information can be utilized in reverse by stochastic sampling to generate diverse samples. However, from a computational perspective, existing score-based models can be efficiently trained only if the forward or the corruption process can be computed in closed form. By using the relationship between the process and layers in a feed-forward network, we derive a backpropagation-based procedure which we call Intermediate Generator Optimization to utilize intermediate iterates of the process with negligible computational overhead. The main advantage of IGO is that it can be incorporated into any standard autoencoder pipeline for the generative task. We analyze the sample complexity properties of IGO to solve downstream tasks like Generative PCA. We show applications of the IGO on two dense predictive tasks viz., image extrapolation, and point cloud denoising. Our experiments indicate that obtaining an ensemble of generators for various time points is possible using first-order methods.

cs.LG↗

arXiv2022

Plasmonic Dimers Enhanced Polarized Single Photon Coupled to Optical Nanowire

We propose a system for guiding plasmon-enhanced polarized single photons into optical nanowire (ONW) guided modes. It is shown that spontaneous emission properties of quantum emitters (QEs) can be strongly enhanced in the presence of gold nanorod dimer (GNRD) leading to the emission of highly polarized and bright single photons. We have calculated that a high Purcell factor of 279, coupling efficiency of 11 %, and degree of polarization (DOP) of single photons is estimated to be as high as 99.57% in the guided modes of ONW by suitably placing a QE on an optimized location of the GNRD system. This proposed hybrid quantum system can be in-line with fiber networks, opening the door for possible quantum information processing and quantum cryptography applications.

physics.optics↗

arXiv2021

Experimental realization of BB84 protocol with different phase gates and SARG04 protocol

Cryptography in the modern era is very important to prevent a cyber attack, as the world tends to be more and more digitalized. Classical cryptographic protocols mainly depend on the mathematical complicacy of encoding functions and the shared key, like RSA protocol in which security depends upon the fact that factoring a big number is a hard problem to the current computers. This means that high computing power can help you crack traditional encryption methods. Quantum machines claim to have this kind of power in many instances. Factorization of big numbers may be possible with Shor's algorithm with quantum machines in considerable time. Apart from this, the main problem is key sharing i.e., how to securely share the key the first time to validate the encryption. Here comes quantum key distribution. Two parties who are interested in communication with each other, create a process, which claims considerable security against an eavesdropper, by encoding and decoding information in quantum states to construct and share a secret key. Quantum key distribution may be done in a variety of ways. This paper begins with experimental verification of the BB84 procedure utilizing four bases (using phase gates) followed by the experimental realization of the SARG04 protocol which was derived from BB84 Protocol to overcome PNS attack. The possibility of a third-party attack and the effect of noise is considered and implemented. The IBM Quantum Experience platform was used for all of the implementations.

quant-ph↗