February 4-6 @ Lumerical’s Booth #5356 (North Hall)
1. The Challenge
Lumerical is hosting a Photonic Inverse Design Competition at Lumerical’s booth #5356 (North Hall) at SPIE Photonics West Conference, February 4-6 at Moscone Conference Center, San Francisco.
Throughout the exhibit, competitors will have an opportunity to apply their best design specifications to a preselected design workflow. All photonic designers are welcome to compete. To assure consistency and fairness, the design and target Figure of Merit will not be revealed prior to the competition. In real time, each competitor’s entry will be created utilizing the power of Lumerical’s Photonic Inverse Design simulated with the industry leading FDTD nanophotonic simulator with the high performance computing capability of Amazon Web Services (AWS). A leader board display in the Lumerical booth will track the standings throughout the exhibition, highlighting the top designs.
The Lumerical Design Competition Champion will be crowned on Thursday, February 6 at 3:00 at the Lumerical Booth. The Champion designer will walk away with bragging rights, a leading design, a MPW ticket to have their design fabricated, and other valuable plus fun prizes to celebrate the victory. All competitors will receive the component that they created for inclusion in their production design. A limited number of leading designs will be chosen for fabrication.
2. Submission Details
Design submission is as easy as entering the following parameters into the iPad provided at the conference booth:
BC Type: Boundry conditions can be clamped, natural or not-a-knot
Control Points: Specified the number of points control points for the spline that defines the y-splitter
Wave Length 1: Specifies the first target wavelength to be split
Delta: Specifies the degree of manufacturing over/under etching for robust design
Wave Length 2: Specifies the second target wavelength to be split
The competition is scored based on the following:
Tmin is the minimum transmission over the bandwidth. Note that there’s a significant cost to allowing your minimum transmission to fall below ~0.3dB!
Verify Your Design
After we automatically generate your design on AWS, you’ll receive an email notification with your mask file (GDS file). KLayout can be used to view your design and is free! Lumerical’s INTERCONNECT seamlessly integrates with KLayout allowing you to quickly simulate your design before tape-out.
4. Inverse Design 101
Inverse Design vs Forward Design
Given a performance goal and design constraints, Lumerical’s inverse design technology automatically and reliably generates optimal designs even with hundreds of free parameters. This is in contrast to traditional forward design approaches which involve a lengthy iterative process of manual design.
Lumerical’s Python based Inverse Design implementation is based on the open-source implementation originally written by Christopher Lalau-Keraly. It’s available on github:
Parametric Shape Based Adjoint Optimization
The parametric shape-based inverse design is based on the efficient adjoint optimization method. Each graduate calculation involves only two FDTD simulations (the forward and adjoint) regardless of the number of parameters.
Measured Results of Devices Designed with Lumopt
Measurements of manufactured designs developed using Lumerical’s Inverse Design have been shown to strongly match simulation results. Below is an example design fabricated by Lukas Chrostowski from the University of British Columbia.
One way to fabricate your own design is to enrol in Dr. Chrostowski’s course which is offered regularly through edX: Silicon Photonics Design, Fabrication, and Data Analysis