Searcharxiv⌕ Search

arXiv subjects

Hamza Kamil

Publications and source records attributed to Hamza Kamil.

2 recordsLinked to original sources

NAP-Wave: A Neural Adaptive Plane-Wave Architecture for High-Frequency Helmholtz Problems

Training neural networks to simulate highly oscillatory wavefields presents optimization challenges, often hindered by spectral bias and poorly conditioned physics-based loss landscapes. To address these challenges and enhance convergence, we propose NAP-Wave, a joint neural architecture and training strategy that expands the target scattered wavefield in an adaptive neural plane-wave basis. The basis functions are extracted from all hidden layers and modulated by trainable plane-wave carriers, embedding the high-frequency oscillatory physics directly into the network's trial space. NAP-Wave is trained on a physics-informed loss built from the Lippmann-Schwinger integral equation, so no differentiation of the network with respect to its input is required. We develop a decoupled optimization strategy in which Adam updates the parameters of the neural plane-wave basis, while the linear combination coefficients are computed exactly by a least-squares (LS) solver. We couple this with an adaptive-depth training strategy in which the network starts shallow and grows progressively layer-by-layer. At each growth step, we avoid the loss spikes typical of dynamic network growth using the LS-computed coefficients, the warm-started parameters, and the preservation of the first- and second-order optimizer moments of the previous stage. Ablation studies confirm that the plane-wave basis, the exact coefficient solve, and the progressive schedule independently accelerate convergence and improve accuracy. Validated on exact-solution benchmarks, NAP-Wave achieves relative L2 errors of order 1%. On highly heterogeneous geophysical benchmarks, it achieves relative errors of 5%, 9%, and 10%, significantly outperforming published PINN methods in both accuracy and convergence time. To ensure reproducibility, the code is available at https://github.com/Hamza-Kamil/NAP-WAVE.

physics.geo-ph↗

Physics-informed neural networks for solving moving interface flow problems using the level set approach

This paper advances the use of physics-informed neural networks (PINNs) architectures to address moving interface problems via the level set method. Originally developed for other PDE-based problems, we particularly leverage PirateNet's features, including causal training, sequence-to-sequence learning, random weight factorization, and Fourier feature embeddings, and tailor them to handle problems with complex interface dynamics. Numerical experiments validate this framework on benchmark problems such as Zalesak's disk rotation and time-reversed vortex flow. We demonstrate that PINNs can efficiently solve level set problems exhibiting significant interface deformation without the need for upwind numerical stabilization, as generally required by classic discretization methods, or additional mass conservation schemes. However, incorporating an Eikonal regularization term in the loss function with an appropriate weight can further enhance results in specific scenarios. Our results indicate that PINNs with the PirateNet architecture surpass conventional PINNs in accuracy, achieving state-of-the-art error rates of $L^2=0.14\%$ for Zalesak's disk and $L^2=0.85 \%$ for the time-reversed vortex flow problem, as compared to reference solutions. Additionally, for a complex two-phase flow dam break problem coupling the level set with the Navier-Stokes equations, we propose a geometric reinitialization method embedded within the sequence-to-sequence training scheme to ensure long-term stability and accurate inference of the level set field.

physics.comp-ph↗