Recent Research on Design Optimization of Wave Propagation in Metamaterials: Implementation-Robust Design and Applications Han Men, Ngoc Cuong Nguyen, Joel Saa-Seoane, Robert Freund, Jaime Peraire AFOSR Optimization and Discrete Mathematics Program Review, April, 2012 7/1/2016 1 Photonic Crystal (PC) Band Gap (BG) Optimization • Forward problem: given a photonic crystal with dielectric property ε(r), we can find its eigenmodes λ’s by solving the governing Maxwell’s equations cast in an eigenvalue problem. • Band gap: a range of prohibitive frequencies over all k’s. The gap-midgap ratio for the mth band gap is defined as: • Inverse problem: find an ε(r), such that its band gap is maximized. ? 7/1/2016 2 SDP Formulation of BG Optimization – P0 Typically, 7/1/2016 3 Parameter Dependence 7/1/2016 4 Change of Decision Variables 7/1/2016 5 Reformulating the problem: P1 We demonstrate the reformulation in the TE case, This reformulation is exact, but non-convex and large-scale. We use a subspace method to reduce the problem size. 7/1/2016 6 Subspace Method 7/1/2016 7 Subspace Method, continued 7/1/2016 8 Subspace Method, continued 7/1/2016 9 Linear Fractional SDP : 7/1/2016 10 Solution Procedure The SDP optimization problems can be solved efficiently using modern interior-point methods [Toh et al. (1999)]. 7/1/2016 11 Need for Fabrication/Implementation Robust Design Consider the optimized PC designs: 7/1/2016 12 Implementation Robustness Paradigm 7/1/2016 13 Implementation Robustness Paradigm, continued 7/1/2016 14 Basic Results 7/1/2016 15 Basic Results: continued 7/1/2016 16 Basic Results: Computation 7/1/2016 17 Computable IR Problems via Special Structure 7/1/2016 18 Computable IR Problems via Special Structure, continued 7/1/2016 19 Computable IR Problems via Special Structure, continued 7/1/2016 20 Implementation Robustness and Photonic Crystal Design 7/1/2016 21 Computable IR Problems via Special Structure, continued 7/1/2016 22 Implementation Robustness Issues when Modeling Eigenvalues 7/1/2016 23 SDP Formulation for Bandgap Optimization 7/1/2016 24 LP Relaxation of the SDP 7/1/2016 25 LP Relaxation of the SDP, continued 7/1/2016 26 LP Relaxation of the SDP, continued linear fractional SDP 7/1/2016 linear fractional program 27 BG optimization: from Nominal to FR Nominal 7/1/2016 FR 28 FR Optimization: Solution Algorithm 7/1/2016 29 Gradient of Parametric LP T. Gal, J. Nedoma, Multiparametric Linear Programming, 1972. 7/1/2016 30 Some results - 1 TE 2nd gap, square lattice, δ = ν = 5% Original: Gap = 65.5% Worst case: Gap = 21.9% 7/1/2016 Fab robust: Gap = 37.2% 5% Manual modification: Gap = 36.1% 31 Some results - 2 TE 1st gap, TM 2nd gap, triangular lattice, δ = ν = 2% Original: Gap = 33.1% Worst case: Gap = 23.5% 7/1/2016 Fab robust: Gap = 29.2% Rand perturb: Gap = 24.9% 32 Some results - 3 TE 1st gap, triangular lattice, δ = ν = 10% Original: Gap = 96.8% Worst case: Gap = 44.5% 7/1/2016 Fab robust: Gap = 80.7% Rand perturb: Gap = 49.8% 33 Bandwidth optimization of Single-polarization Single-mode Photonic Crystal Fiber Applications of photonic crystals based on index guiding 1 2 1. Index-guiding photonic fiber (PCF). [T. A. Birks et al., Opt. Lett. 22, 961, 1997.] 2. Photonic crystal slab. [S. Assefa et al, Appl Phys Lett 85 25, 2004.] Design problem Guided Mode Unguided Mode 7/1/2016 1. Maximize the dimensionless quantity: bandwidth 2. Enforce the following two constraints 34 Single-polarization Single-mode Photonic Crystal Fiber: Optimal structures Initial structure UNDAMENTAL SFECOND MODE MODE After cladding optimization Seconded mode is attenuated Fundamental guided mode is localized After core + cladding optimization Confinement is enhanced. 35 Bandgap optimization of 3D photonic crystal Applications of photonic crystals based on band gaps 1. Hollow-core photonic band gap fiber. [Mangan, et al., OFC 2004 PDP24.] 2. Photonic crystal switch. [Beggs, et al., IEEE Photonics Tech. Lett. 21, 2009.] 3. Photonic crystal splitter. [Communication Photonics Group, ETH.] Joannopoulos et al, 2007 36 Binary optimization of Metamaterials • Metamaterials are artificially engineered materials that have a very special property due to its structure. • We focus on wave phenomena. Applications: 1. Shock Dispersion 4. Superlensing 2. Negative Poisson 5. Water front control 6. Nonlinear earplugs 3. Cloaking 3 1 5 Spadoni et al. 2 4 Light Fluid Dynamics 6 Elasticity 7/1/2016 Acoustics 37 Conclusions • Solved the photonic crystal bandgap optimization with an LP relaxation of SDP formulation, enhanced by delayed constraint generation • Developed a framework for Implementation Robustness, and extended it to the Fabrication Robustness of photonic 2D crystal bandgap optimization. • The FR optimal structures provide a starting point to modify within a given allowance and fabricate an easier structure, while knowing the worst case scenario (or a lower bound of the bandgap). • We have extended our methodology to PCFs and working on full 3D • We are looking at new metamaterial design applications 7/1/2016 40