Computational Physicist | Electromagnetics and Multiphysics
Abu Dhabi, UAE
BS (NTNU-Physics), PhD (NTU-Physics)
I am a computational physicist with a strong interest in turning physical models into practical simulation workflows. My work spans first-principles calculations, computational electromagnetics, and multiphysics structural modeling for heterogeneous and multiscale systems.
I am particularly interested in combining analytical modeling, numerical simulation, and experimental measurements to relate material-level mechanisms to measurable device and structural behavior.
These selected research areas highlight the physical problems, modeling approaches, and representative systems that connect my work in first-principles simulation, multiscale materials modeling, optics, and multiphysics analysis.
This work combines local probe measurements and first-principles analysis to study graphene-coated interfaces and other low-dimensional material systems. The emphasis is on how interfacial structure and surface energetics influence measurable behavior, which is directly relevant to functional heterostructures, surface engineering, and materials characterization. The related work has been published in Nanoscale, which could be accessed here.
The figure below compares local surface-energy behavior for graphene-coated substrates prepared by different routes.
This research develops first-principles routes for predicting wettability, adhesion, and interfacial energetics in solid–liquid and multiphase systems. It connects atomistic calculations to engineering observables such as contact angle and spreading behavior, and is relevant to surface design, interfacial materials, and physics-based materials screening. The related work has been published in Journal of Physical Chemistry Letters, which could be accessed here.
The figure below illustrates an atomistic workflow for predicting macroscopic wetting behavior from surface-specific calculations.
This work combines device design, thermal transport analysis, and multiphysics simulation to understand interfacial solar vapor generation under realistic operating conditions. It reflects a broader interest in linking material properties, heat transfer, and system-level performance in engineered energy devices. The related work has been published in Joule, which could be accessed here.
The figure below compares measured and simulated temperature fields for the vapor generator under different illumination conditions.
This research uses electromagnetic simulation, nanocomposite design, and experiment to develop broadband optical absorbers with scalable thin-film architectures. The work sits at the interface of photonic materials engineering and structure–property design, and is closely aligned with computationally guided materials discovery for functional coatings and optical materials. The related work has been published in Advanced Optical Materials, which could be accessed here.
The figure below shows how embedded metallic nanoparticles strengthen light–matter interaction in nanocomposite absorber films.
This work focuses on wave-based design of ultrathin absorbing structures by combining thin-film interference with localized plasmonic effects. It highlights a physics-driven route for engineering optical response through geometry, materials selection, and simulation, which aligns well with advanced materials design and device-oriented computational research. The related work has been published in Advanced Optical Materials, which could be accessed here.
The figure below illustrates an ultrathin broadband absorber in which pore-shaped metallic features activate localized plasmonic enhancement.
The following are the full list of publications
These notes sit between the publication list and the teaching modules. They emphasize how physical models are formulated, validated, and translated into defensible computational workflows for materials, devices, and engineered structures.
Notes on homogenization, Maxwell–Garnett theory, Bergman spectral representation, and the interpretation of microstructure-dependent effective response in heterogeneous media.
This stream also includes revisiting classical formulations beyond the basic dipolar Maxwell–Garnett limit, including PRB 1992-type treatments of interaction effects, spectral response, and the physically meaningful reconstruction of published composite-medium results.
Notes on DFT-based workflows, descriptor selection, structure–property mapping, and physically grounded strategies for computational materials screening and accelerated materials discovery.
Methodological notes on combining simulation, optimization, and machine-learning-assisted search for functional materials and engineered structures, with emphasis on interpretable design logic rather than black-box prediction alone.
Notes linking analytical models, finite element simulation, and engineering observables across structural, thermal, electromagnetic, and heterogeneous material systems, with attention to scale bridging and physically meaningful performance metrics.
Notes on model validation, convergence assessment, analytical benchmarking, code verification, and physics-based interpretation across electromagnetic, mechanical, and atomistic simulation workflows.
Optical material models
This section provides a searchable database of compact optical material models for time-domain electromagnetic simulation. The current visible–near-IR release is organized around complex-conjugate pole-residue rational-pole models (CCPR/RP) and critical-point (CP) extensions of the same pole-residue framework.
RP provides the general complex pole-residue representation, while CP introduces critical-point-inspired structure for optical spectra with interband-like features. In both cases, the final fitted parameters are exported as PoleResidue JSON files following the Tidy3D-compatible file convention for easy implementation and are used directly by the centered-ADE FDTD update.
In the PoleResidue convention used by the exported JSON files, the CCPR/RP model is written as a complex-conjugate pole-residue expansion,
The complex-conjugate pairing gives a real time-domain response, while the positive decay convention is converted to the centered-ADE update coefficients used in FDTD.
The CP model is used as a positive-frequency critical-point fitting ansatz for interband-like optical features,
Here Ωℓ, Γℓ, Aℓ, φℓ, and pℓ denote the critical-point energy, broadening, strength, phase, and line-shape exponent. This CP expression is used for fitting the optical response; the fractional-power CP term is not stored directly as the FDTD update model.
For the released JSON files, each accepted CP fit is converted to a finite CCPR/PoleResidue representation,
Therefore, users download and implement the converted CCPR/PoleResidue JSON parameters, and the FDTD implementation remains a centered-ADE PoleResidue update.
The fitting workflow does not select models by RMS error alone. Exported models are screened for causal pole placement, passive dielectric response on the fitted wavelength grid, valid PoleResidue JSON export, available model-response curves, and numerical readiness for the centered-ADE update at CFL = 0.75.
The figures below summarize the relative complex-permittivity RMS error as a function of pole-pair order N on common optical records. The reference curve corresponds to an external on-demand pole-residue fitter, while CCPR and CP denote the database fitting families. Increasing N generally reduces the median error of the released CCPR and CP candidates, but the final recommendation is selected using both fitting accuracy and physical/numerical readiness.
Use this database when you need an already-audited dispersive material model for FDTD within the fitted visible–near-IR interval. Each record is fitted independently over the wavelength range supported by its own experimental data, usually within 0.3–1.1 µm or a subset of that range.
These models should not be treated as universal optical constants outside the fitted interval. Extrapolation can be inaccurate even when a model is causal, passive, and usable in the centered-ADE update. Users should inspect the online n/k and permittivity plots before using a model in production simulations.
| Candidate model rows | 6075 |
|---|---|
| Exported centered-ADE-ready model files | 3929 |
| Records with final recommendation | 396 / 405 |
| Records without recommended export | 9 / 405 |
| Full fitted 0.3–1.1 µm records | 279 / 405 |
| Partial-window records | 126 / 405 |
Experimental optical constants are converted to εexp(λ) = [n(λ) + i k(λ)]2 and compared with the fitted εfit(λ) on the same fitting grid. The reported error is ε-RMS = sqrt(mean(|εfit − εexp|2 / (|εexp|2 + 10−12))).
| Recommended ε-RMS ≤ 10−3 | 59 / 396 |
|---|---|
| Recommended ε-RMS ≤ 5×10−3 | 241 / 396 |
| Recommended ε-RMS ≤ 10−2 | 281 / 396 |
| Recommended ε-RMS > 5×10−2 | 39 / 396 |
These statistics use the final recommended model for each recommended record, so the denominator is 396 rather than the 3929 exported model files.
Questions, corrections, or suggestions: jinyoulu@ntu.edu.tw.
Thank you for all the supports during my postdoc career:
Thank you for the supports during my PhD:
Thank you for the supports during my career at TSMC:
This section collects selected teaching modules and technical tutorials in computational mechanics, computational electromagnetics, scientific computing, and materials modeling.
Computational Structural Mechanics
Analytical modeling, CLT, COMSOL simulation, and simulation-oriented structural notes.
Computational Electromagnetics
TMM, FDTD, wave propagation, scattering, and simulation-oriented electromagnetic notes.
Multilayer Transfer Matrix Calculations
Fresnel coefficients, transfer matrices, multilayer reflectance and transmittance.
CUDA Tutorial for Scientific Computing
GPU computing, kernel design, memory access, and performance-oriented numerical workflows.
QuantumEspresso Si Bandstructure
Plane-wave DFT workflow, Brillouin-zone path, and basic electronic-structure analysis.
Filmmetric F40 Equipment Measurements
Thickness measurement, optical fitting, and thin-film characterization basics.