arXiv subjects

Steven H. Frankel

Publications and source records attributed to Steven H. Frankel.

At least 19 recordsLinked to original sources

arXiv2026

Supersonic jet impingement on concave surfaces

The aeroacoustic resonance of round supersonic jets impinging on concave surfaces is investigated using compressible large-eddy simulations, vortex-sheet modelling, and Powell's feedback-loop analysis. The choked jets operate at an ideally expanded Mach number of $1.56$ and a Reynolds number of $6\times10^4$. Six geometries are considered: two flat plates at $L/D=2.08$ and $2.58$, where $L$ is the nozzle-to-wall distance and $D$ the nozzle exit diameter, and four Gaussian concave surfaces of fixed depth and indentation spread $\sigma\in\{0.4,0.8,1.6,4.0\}$. As the indentation narrows, the primary-tone amplitude increases by up to $23\,\mathrm{dB}$ relative to the flat-wall reference at $L/D=2.6$, together with larger wall-pressure fluctuations and moments. A Powell-Tam source-transfer budget attributes this amplification to increased Mach-disk source amplitude and more efficient return of the upstream feedback wave to the nozzle. The stronger upstream-propagating waves are consistent with acoustic focusing by the concave wall. For the helical cases, the measured frequencies and radial eigenfunctions agree closely with the guided jet mode predicted by the vortex-sheet model, supporting its role in closing the upstream feedback path. The same selection is recovered for concave and flat walls alike, so this tone is governed by the shear-layer profile of the equivalent ideally expanded jet rather than by the wall geometry. The axisymmetric frequencies, by contrast, coincide with no guided-mode branch and appear instead to follow Powell's classical loop-length criterion. The results identify distinct frequency-selection mechanisms for helical and axisymmetric screech and demonstrate that wall curvature provides effective control of screech amplitude and surface loading.

physics.flu-dyn↗

arXiv2026

Dynamic masking for boundary-aware velocity reconstruction in volumetric particle tracking with moving solids

Volumetric particle tracking velocimetry (PTV) produces scattered Lagrangian tracks that must be reconstructed on an Eulerian grid before velocity gradients, pressure, or hydrodynamic loads can be evaluated. This step is usually performed on a domain treated as entirely fluid. When a solid body lies within the measurement volume, its surface kinematics are not imposed and the reconstruction is weakest in the steep-gradient region next to the body. We introduce LE-DM (Lagrangian-to-Eulerian reconstruction with Dynamic Masking), a constrained reconstruction framework for moving solid boundaries. A time-dependent signed-distance function classifies grid nodes as open fluid, boundary shell, or solid interior. The particle data, incompressibility constraint, prescribed surface velocity, and regularization terms are then assembled on the masked domain within a single solve. The method requires only a signed-distance field and a surface velocity, allowing stationary walls, translating, rotating, multiple, and deforming bodies to be represented in the same formulation. LE-DM is assessed using an analytical oscillating sphere, synthetic tracks from a CFD rising-sphere simulation, and a refractive-index-matched tomographic-PTV experiment on a freely rising sphere. The surface kinematics are enforced to solver tolerance, while the bulk reconstruction remains unchanged where no body is present. In the analytical case, the first-cell error is reduced from 14\% to 3\% of the body speed. In the experiment, LE-DM recovers the independently measured surface velocity, whereas an all-fluid reconstruction does not. The result is a divergence-free, boundary-consistent velocity field for pressure and force estimation.

physics.flu-dyn↗

arXiv2026

Hybrid Quantum-Classical PINNs for Scientific Computing: A Multi-GPU Open-Source Framework

We present QPINNACLE, an open-source computational framework for physics-informed neural networks (PINNs) that integrates modern training strategies, multi-GPU acceleration, and hybrid quantum-classical architectures within a unified modular workflow. The framework enables systematic evaluation of PINN performance across benchmark problems including 1D hyperbolic conservation laws, incompressible flows, and electromagnetic wave propagation. It supports a range of architectural and training enhancements, including Fourier feature embeddings, random weight factorization, strict boundary condition enforcement, adaptive loss balancing, curriculum training, and second-order optimization strategies, with extensibility to additional methods. We provide a comprehensive benchmark study quantifying the impact of these methods on convergence, accuracy, and computational cost, and analyze distributed data parallel scaling in terms of runtime and memory efficiency. In addition, we extend the framework to hybrid quantum-classical PINNs and derive a formal estimate for circuit-evaluation complexity under parameter-shift differentiation. Results highlight the sensitivity of PINNs to architectural and training choices, confirm their high computational cost relative to classical solvers, and identify regimes where hybrid quantum models offer improved parameter efficiency. QPINNACLE provides a foundation for benchmarking physics-informed learning methods and guiding future developments through quantitative assessment of their trade-offs.

cs.LG↗

arXiv2026

Gaussian Field Representations for Turbulent Flow: Compression, Scale Separation, and Physical Fidelity

Representing turbulent flow fields in a compact yet physically faithful form remains a central challenge in computational fluid dynamics. We propose a continuous parametric representation based on localized Gaussian primitives, in which the velocity field is modeled as a superposition of kernels with learnable positions, amplitudes, and scales. This formulation yields a compact, grid-independent encoding while enabling evaluation of derived quantities such as vorticity and enstrophy. The approach is assessed on three-dimensional Taylor-Green vortex fields spanning stages from smooth flow to fully developed turbulence. We quantify the compression-accuracy trade-off using both primary variables and derivative-sensitive diagnostics. The baseline isotropic formulation achieves high velocity accuracy at compression ratios exceeding 1e3-1e4, but exhibits substantial enstrophy degradation due to loss of small-scale structure. To address this limitation, we investigate structure-aware extensions including adaptive placement, multi-resolution kernels, and anisotropic Gaussians. The anisotropic formulation provides the most consistent improvement, better aligning with elongated vortical structures and recovering intermediate- and high-wavenumber content, while other strategies yield modest gains. A compact-support Beta basis improves enstrophy in some cases but introduces localized artifacts. Overall, the results indicate that the main limitation of baseline Gaussian representations lies in geometric expressiveness rather than parameter count. The proposed framework provides a compact, interpretable, and continuous representation of turbulent flows, and establishes a foundation for structure-aware and physics-informed flow compression.

physics.flu-dyn↗

arXiv2026

Shock propagation through a local constriction

The interaction of a shock wave with a localized constriction in a straight conduit is investigated by systematically varying the blockage ratio in the range 0.35-0.75, the normalized constriction length in the range 0.25-2, and the incident Mach numbers of 1.4 and 1.8. Abrupt rectangular constrictions and smoothly contoured sinusoidal constrictions are considered, as they provide two limiting configurations. Validated Large-eddy simulations resolve both the transient start-up dynamics and the subsequent propagation of reflected and transmitted shock waves. The results show that, for rectangular constrictions, the reflected shock strength depends primarily on the blockage ratio and is largely independent of length, whereas the transmitted shock exhibits measurable sensitivity to constriction length. In contrast, sinusoidal constrictions display a strong coupling between blockage and length, with the reflection process governed by the local contour slope and evolving reflection topology. The start-up process within the constriction occurs over time scales one to two orders of magnitude longer than the shock passage time and is characterized by a sequence of reflection, separation, and flow reorganization events that determine the eventual steady shock configuration. At later times, the reflected shock Mach number scales linearly with blockage ratio, while the transmitted shock strength decreases monotonically with increasing blockage. Based on these trends, semi-empirical models are developed to predict the strengths of both reflected and transmitted shocks across the parameter space considered. These results provide a unified framework for understanding and predicting shock propagation in conduits with localized geometric variations, with direct relevance to compressible internal flows in engineering and natural systems.

physics.flu-dyn↗

arXiv2025

Improving Accuracy and Efficiency of Implicit Neural Representations: Making SIREN a WINNER

We identify and address a fundamental limitation of sinusoidal representation networks (SIRENs), a class of implicit neural representations. SIRENs Sitzmann et al. (2020), when not initialized appropriately, can struggle at fitting signals that fall outside their frequency support. In extreme cases, when the network's frequency support misaligns with the target spectrum, a 'spectral bottleneck' phenomenon is observed, where the model yields to a near-zero output and fails to recover even the frequency components that are within its representational capacity. To overcome this, we propose WINNER - Weight Initialization with Noise for Neural Representations. WINNER perturbs uniformly initialized weights of base SIREN with Gaussian noise - whose noise scales are adaptively determined by the spectral centroid of the target signal. Similar to random Fourier embeddings, this mitigates 'spectral bias' but without introducing additional trainable parameters. Our method achieves state-of-the-art audio fitting and significant gains in image and 3D shape fitting tasks over base SIREN. Beyond signal fitting, WINNER suggests new avenues in adaptive, target-aware initialization strategies for optimizing deep neural network training. For code and data visit cfdlabtechnion.github.io/siren_square/.

cs.CV↗

arXiv2025

Spectral Bottleneck in Sinusoidal Representation Networks: Noise is All You Need

This work identifies and attempts to address a fundamental limitation of implicit neural representations with sinusoidal activation. The fitting error of SIRENs is highly sensitive to the target frequency content and to the choice of initialization. In extreme cases, this sensitivity leads to a spectral bottleneck that can result in a zero-valued output. This phenomenon is characterized by analyzing the evolution of activation spectra and the empirical neural tangent kernel (NTK) during the training process. An unfavorable distribution of energy across frequency modes was noted to give rise to this failure mode. Furthermore, the effect of Gaussian perturbations applied to the baseline uniformly initialized weights is examined, showing how these perturbations influence activation spectra and the NTK eigenbasis of SIREN. Overall, initialization emerges as a central factor governing the evolution of SIRENs, indicating the need for adaptive, target-aware strategies as the target length increases and fine-scale detail becomes essential. The proposed weight initialization scheme (WINNER) represents a simple ad hoc step in this direction and demonstrates that fitting accuracy can be significantly improved by modifying the spectral profile of network activations through a target-aware initialization. The approach achieves state-of-the-art performance on audio fitting tasks and yields notable improvements in image fitting tasks.

eess.AS↗

arXiv2025

Quantum Singular Value Transformation for Solving the Time-Dependent Maxwell's Equations

This work presents a quantum algorithm for solving linear systems of equations of the form $\mathbf{A}{\frac{\mathbf{\partial f}}{\mathbf{\partial x}}} = \mathbf{B}\mathbf{f}$, based on the Quantum Singular Value Transformation (QSVT). The algorithm uses block-encoding of $A$ and applies an 21st-degree polynomial approximation to the inverse function $f(x) = 1/x$, enabling relatively shallow quantum circuits implemented on 9 qubits, including two ancilla qubits, corresponding to a grid size of 128 points. Phase angles for the QSVT circuit were optimized classically using the Adagrad gradient-based method over 100 iterations to minimize the solution cost. This approach was simulated in PennyLane and applied to solve a 1D benchmark case of Maxwell's equations in free space, with a Gaussian pulse as the initial condition, where the quantum-computed solution showed high fidelity of more than 99.9% when compared to the normalized classical solution. Results demonstrate the potential of QSVT-based linear solvers on simulators with full quantum state access. However, practical hardware implementations face challenges because accessing the complete quantum state is infeasible. This limitation restricts applicability to cases where only $O({poly}(n))$ observables are needed. These findings highlight both the promise and current limitations of using quantum algorithms, such as QSVT, to solve linear systems of equations, and they point to the need for the development of measurement-efficient algorithms for near-term quantum devices.

quant-ph↗

arXiv2025

Quantum Physics-Informed Neural Networks for Maxwell's Equations: Circuit Design, "Black Hole" Barren Plateaus Mitigation, and GPU Acceleration

Physics-Informed Neural Networks (PINNs) have emerged as a promising approach for solving partial differential equations (PDEs) by embedding the governing physics into the loss function associated with a deep neural network. In this work, a Quantum PINNs (QPINN) framework is proposed to solve two-dimensional (2D) time-dependent Maxwell's equations. Our approach utilizes a parametrized quantum circuit in conjunction with the classical neural network architecture and enforces physical laws, including a global energy conservation principle, during training. A quantum simulation library, TorQ, was developed to efficiently compute circuit outputs and derivatives by leveraging GPU acceleration based on PyTorch, enabling end-to-end training of the QPINN. The method was evaluated on two 2D electromagnetic wave propagation problems: one in free space (vacuum) and the other has an added dielectric medium. Multiple quantum circuit ans\"atze, input scales, and an added loss term were compared in a thorough ablation study. Furthermore, recent techniques to enhance PINN convergence, including random Fourier feature embeddings and adaptive time weighting, have been incorporated. Our results demonstrate that the QPINN achieves accuracy comparable to, and even greater than, the classical PINN baseline, while using a significantly smaller number of trainable parameters. This study also shows that adding an energy conservation term to the loss stabilizes training and improves the physical fidelity of the solution in the lossless free-space case. This added term helps mitigate a new kind of barren plateau (BP) related phenomenon - ``black hole'' (BH) loss landscape for the quantum experiments in that scenario. By optimizing the quantum-circuit ansatz and embedding energy-conservation constraints, our QPINN achieves up to a 19% higher accuracy on 2D Maxwell benchmark problems compared to a classical PINN.

quant-ph↗

arXiv2025

Deep Learning vs. Black-Scholes: Option Pricing Performance on Brazilian Petrobras Stocks

This paper explores the use of deep residual networks for pricing European options on Petrobras, one of the world's largest oil and gas producers, and compares its performance with the Black-Scholes (BS) model. Using eight years of historical data from B3 (Brazilian Stock Exchange) collected via web scraping, a deep learning model was trained using a custom built hybrid loss function that incorporates market data and analytical pricing. The data for training and testing were drawn between the period spanning November 2016 to January 2025, using an 80-20 train-test split. The test set consisted of data from the final three months: November, December, and January 2025. The deep residual network model achieved a 64.3\% reduction in the mean absolute error for the 3-19 BRL (Brazilian Real) range when compared to the Black-Scholes model on the test set. Furthermore, unlike the Black-Scholes solution, which tends to decrease its accuracy for longer periods of time, the deep learning model performed accurately for longer expiration periods. These findings highlight the potential of deep learning in financial modeling, with future work focusing on specialized models for different price ranges.

q-fin.ST↗

arXiv2025

PINNs for Solving Unsteady Maxwell's Equations: Convergence Issues and Comparative Assessment with Compact Schemes

Physics-Informed Neural Networks (PINNs) have recently emerged as a promising alternative for solving partial differential equations, offering a mesh-free framework that incorporates physical laws directly into the learning process. In this study, we explore the application of PINNs for solving unsteady Maxwell's equations and compare their performance with two established numerical methods: the Finite-Difference Time-Domain (FDTD) method and a compact Pade scheme with filtering. Three benchmark problems are considered, ranging from 1D free-space wave propagation to 2D Gaussian pulses in periodic and dielectric media. We assess the effectiveness of convergence-enhancing strategies for PINNs, including random Fourier features, spatio-temporal periodicity, and temporal causality training. An ablation study highlights that architectural choices must align with the underlying physics. Additionally, we employ a Neural Tangent Kernel framework to examine the spatio-temporal convergence behavior of PINNs. Results show that convergence rates correlate with error over time but not in space, revealing a limitation in how training dynamics allocate learning effort. Overall, this study demonstrates that PINNs, when properly configured, can match or surpass traditional solvers in accuracy and flexibility, though challenges remain in addressing spatial inhomogeneity and adapting training to localized complexity.

physics.comp-ph↗

arXiv2025

Quantum Unitary Matrix Representation of Lattice Boltzmann Model for Low Reynolds Fluid Flow Simulation

We propose a quantum algorithm for the Lattice Boltzmann (LB) method to simulate fluid flows in the low Reynolds number regime. First, we encode the particle distribution functions (PDFs) as probability amplitudes of the quantum state and demonstrate the need to control the state of the ancilla qubit during the initial state preparation. Second, we express the LB algorithm as a matrix-vector product by neglecting the quadratic non-linearity in the equilibrium distribution function, wherein the vector represents the PDFs, and the matrix represents the collision and streaming operators. Third, we employ classical singular value decomposition (SVD) to decompose the non-unitary collision and streaming operators into a product of unitary matrices. Finally, we show the importance of having a Hadamard gate between the collision and the streaming operations. Our approach has been tested on linear/linearized flow problems such as the advection-diffusion of a Gaussian hill, Poiseuille flow, Couette flow, and lid-driven cavity problems. We provide counts for two-qubit controlled-NOT (CNOT) and single-qubit U gates for test cases involving 9 to 12 qubits, with grid sizes ranging from 24 to 216 points. While the gate count aligns closely with theoretical limits, the high number of two-qubit gates on the order of $10^7$ necessitates careful attention to circuit synthesis.

quant-ph↗

arXiv2024

Decomposition of Nonlinear Collision Operator in Quantum Lattice Boltzmann Algorithm

We propose a quantum algorithm to tackle the quadratic nonlinearity in the Lattice Boltzmann (LB) collision operator. The key idea is to build the quantum gates based on the particle distribution functions (PDF) within the coherence time for qubits. Thus, both the operator and a state vector are linear functions of PDFs, and upon quantum state evolution, the resulting PDFs will have quadraticity. To this end, we decompose the collision operator for a $DmQn$ lattice model into a product of $2(n+1)$ operators, where $n$ is the number of lattice velocity directions. After decomposition, the $(n+1)$ operators with constant entries remain unchanged throughout the simulation, whereas the remaining $(n+1)$ will be built based on the statevector of the previous time step. Also, we show that such a decomposition is not unique. Compared to the second-order Carleman-linearized LB, the present approach reduces the circuit width by half and circuit depth by exponential order. The proposed algorithm has been verified through the one-dimensional flow discontinuity and two-dimensional Kolmogrov-like flow test cases.

quant-ph↗

arXiv2024

High Resolution Optimized High-Order Schemes for Discretization of Non-Linear Straight and Mixed Second Derivative Terms

In this paper, we propose a new set of midpoint-based high-order discretization schemes for computing straight and mixed nonlinear second derivative terms that appear in the compressible Navier-Stokes equations. Firstly, we detail a set of conventional fourth and sixth-order baseline schemes that utilize central midpoint derivatives for the calculation of second derivatives terms. To enhance the spectral properties of the baseline schemes, an optimization procedure is proposed that adjusts the order and truncation error of the midpoint derivative approximation while still constraining the same overall stencil width and scheme order. A new filter penalty term is introduced into the midpoint derivative calculation to help achieve high wavenumber accuracy and high-frequency damping in the mixed derivative discretization. Fourier analysis performed on the both straight and mixed second derivative terms show high spectral efficiency and minimal numerical viscosity with no odd-even decoupling effect. Numerical validation of the resulting optimized schemes is performed through various benchmark test cases assessing their theoretical order of accuracy and solution resolution. The results highlight that the present optimized schemes efficiently utilize the inherent viscosity of the governing equations to achieve improved simulation stability - a feature attributed to their superior spectral resolution in the high wavenumber range. The method is also tested and applied to non-uniform structured meshes in curvilinear coordinates, employing a supersonic impinging jet test case.

math.NA↗

arXiv2024

Centralized Gradient-Based Reconstruction for Wall Modelled Large Eddy Simulations of Hypersonic Boundary Layer Transition

In this study, we introduce a robust central Gradient-Based Reconstruction (GBR) scheme for the compressible Navier-Stokes equations. The method leverages transformation to characteristic space, allowing selective treatment of waves from the compressible Euler equations. By averaging left- and right-biased state interpolations, a central scheme is achieved for all but the acoustic waves, which require upwinding for stability. Distinct differences were observed between transformations using either primitive or conservative variables. We evaluated the method's robustness and superiority using benchmark problems, including the two-dimensional shock entropy problem, two-dimensional viscous shock tube, and three-dimensional inviscid Taylor-Green vortex. Subsequently, we assessed the method in the context of Wall Modelled Large Eddy Simulations (WMLES), where coarse grids are used to reduce computational cost but also introduce substantial numerical dissipation. Using WMLES, we simulated oblique shock impingement on a Mach 6 disturbed boundary layer and a Mach 7.7 flow over a $15^{\circ}$ compression ramp. Our findings reveal that: 1) transformation to characteristic space using conservative variables leads to more accurate results; 2) minimizing numerical dissipation through centralized interpolation is crucial. In the compression ramp case, boundary layer separation was shifted slightly upstream, and there was an over-prediction of wall heating, likely attributable to the equilibrium-assuming wall model. Overall, this work showcases the method's potential in accurately capturing complex flow dynamics with reduced numerical dissipation.

physics.flu-dyn↗

arXiv2023

On the Application of Gradient Based Reconstruction for Flow Simulations on Generalized Curvilinear and Dynamic Mesh Domains

Accurate high-speed flow simulations of practical interest require numerical methods with high-resolution properties. In this paper, we present an extension and demonstration of the high-accuracy Gradient-based reconstruction and $α$-damping schemes introduced by Chamarthi (2022) [1] for simulating high-speed flows in generalized curvilinear and dynamic mesh domains with the freestream preservation property. In the first part of this paper, the algorithms are detailed within the generalized curvilinear coordinate framework, with a focus on demonstration through stationary and dynamic mesh test cases. It has been shown both theoretically and through the use of test cases that the conservative metrics, including their interpolation to cell interfaces, must be numerically computed using a central scheme that is consistent with the inviscid flux algorithm to achieve the freestream preservation property. The second part of the paper illustrates the efficacy of the algorithm in simulating supersonic jet screech by displaying its capability to capture the screech tones and accurately characterize the unsteady lateral flapping mode of a Mach 1.35 under-expanded supersonic jet, in contrast to the WENO-Z scheme which fails to do so at the same grid resolution. In the final part of the paper, the parallelizability of the schemes on GPU architectures is demonstrated and performance metrics are evaluated. A significant speedup of over $200 \times$ (compared to a single core CPU) and a reduction in simulation completion time to 34.5 hours per simulation were achieved for the supersonic jet noise case at a grid resolution of 13 million cells.

physics.flu-dyn↗

arXiv2023

PANSATZ: Pulse-based Ansatz for Variational Quantum Algorithms

We develop and implement a novel pulse-based ansatz, which we call PANSATZ, for more efficient and accurate implementations of variational quantum algorithms (VQAs) on today's noisy intermediate-scale quantum (NISQ) computers. Our approach is applied to quantum chemistry. Specifically, finding the ground-state energy associated with the electron configuration problem, using the variational quantum eigensolver (VQE) algorithm for several molecules. We manage to achieve chemical accuracy both in simulation for several molecules and on one of IBM's NISQ devices for the $H_2$ molecule in the STO-3G basis. Our results are compared to a gate-based ansatz and show significant latency reduction - up to $7\times$ shorter ansatz schedules. We also show that this ansatz has structured adaptivity to the entanglement level required by the problem.

quant-ph↗

arXiv2023

Implicit gradients based conservative numerical scheme for compressible flows

This paper introduces a novel approach to compute the numerical fluxes at the cell boundaries for a cell-centered conservative numerical scheme. Explicit gradients used in deriving the reconstruction polynomials are replaced by high-order gradients computed by compact finite differences, referred to as implicit gradients in this paper. A problem-independent shock capturing approach via Boundary Variation Diminishing (BVD) algorithm is used to suppress oscillations for the simulation of flows with shocks and material interfaces. Several numerical test cases are carried out to verify the proposed method's capability using the implicit gradient method for compressible flows.

math.NA↗