UCDelhiMarch 2015 FINITEDIFFERENCE TIMEDOMAIN COMPUTATIONS IN NANOPHOTONICS Prof

Mesoscopic Optics � ‘Meso’: greek for ‘middle’, for `intermediate-sized’, between micro and macro �

Mesoscopic Optics of disordered systems Active systems Passive systems Diffusion of light increasing disorder

Mesoscopic Optics of disordered systems Active systems Passive systems Diffusion of light Weak localization

Computations � � � Photon Transport Monte Carlo Simulation MEEP: Finite difference time domain

Finite-difference time-domain Electromagnetic ‘wave’ propagation: Maxwell’s equations:

Computations Maxwell’s equations: In one dimension: Yee Cell (1966) Discretized:

Computations Show movie (s)! Resonance! Random Resonance!

Challenges! � Computational resources! � Memory, Storage space, Speed. � Several arrays are generated,

Some thoughts… � FDTD/Maxwell is now a universal tool for optics/photonics � Enhanced capability

Slides: 13

Download presentation

UC-Delhi-March 2015 FINITE-DIFFERENCE TIME-DOMAIN COMPUTATIONS IN NANOPHOTONICS Prof Sushil Mujumdar Nano-optics and Mesoscopic Optics Laboratory Tata Institute for Fundamental Research, Mumbai, India. mujumdar@tifr. res. in http: //www. tifr. res. in/~mujumdar

Mesoscopic Optics � ‘Meso’: greek for ‘middle’, for `intermediate-sized’, between micro and macro � Origin from Condensed Matter Physics, where the sample is smaller than inelastic lengthscales. � Aim: Effect of disorder on electron wave transport studied optically. � Advantages due to well-defined energies, transport timescales, measurement capabilities etc! � Scatterers for light need to be sub-wavelength, hence… Nano-Optics

Mesoscopic Optics of disordered systems Active systems Passive systems Diffusion of light increasing disorder Weak localization Anderson localization

Mesoscopic Optics of disordered systems Active systems Passive systems Diffusion of light Weak localization Synergy of amplification and multiple scattering Random lasers Anderson localization

Computations � � � Photon Transport Monte Carlo Simulation MEEP: Finite difference time domain (FDTD) LUMERICAL: FDTD with saturable gain for lasing studies. MPB: Frequency-domain eigensolver COMSOL: Finite-element method for plasmonic structures

Finite-difference time-domain Electromagnetic ‘wave’ propagation: Maxwell’s equations:

Computations Maxwell’s equations: In one dimension: Yee Cell (1966) Discretized:

Computations Show movie (s)! Resonance! Random Resonance!

Challenges! � Computational resources! � Memory, Storage space, Speed. � Several arrays are generated, used and sometimes required to be stored. Anisotropy, gain, etc… � � � High Performance Cluster (32/96 cores, 128 GB memory, 5 Teraflops) In the lab: Dell Precision Workstation (Six cores, 24 GB memory) Isn’t this sufficient? ? ?

Why want more? Real samples!

Some thoughts… � FDTD/Maxwell is now a universal tool for optics/photonics � Enhanced capability is desired, but dedicated efforts/infrastructure is unavailable. � Theory X Experiment disconnect is closing. � Optics Photonics is hugely industry-applicable research. � Maybe other tools exist in other fields of experimental research. � One thinkable idea: To put one/some of these tools in parallel format on a network/grid and invite people to use it. Thank You!