From inverse design to implementation of practical (quantum) photonics (Conference Presentation)

The convergence of artificial intelligence and machine learning with material science holds significant promise to rapidly accelerate development timelines of new high-performance polymeric materials. Within this context, we report an inverse design strategy for polycarbonate and polyester discovery based on a recommendation system that proposes polymerization experiments that are likely to produce materials with targeted properties. Following recommendations of the system driven by the historical ring-opening polymerization results, we carried out experiments targeting specific ranges of monomer conversion and dispersity of the polymers obtained from cyclic lactones and carbonates. The results of the experiments were in close agreement with the recommendation targets with few false negatives or positives obtained for each class.<br>

Download Full-text

Inverse design of 3D curved ducts using a 3D-upgraded ball-spine algorithm

Inverse Problems in Science and Engineering ◽

10.1080/17415977.2021.1894143 ◽

2021 ◽

pp. 1-35

Author(s):

Atefeh Kariminia ◽

Mahdi Nili-Ahmadabadi ◽

Kyung Chun Kim

Keyword(s):

Inverse Design ◽

Curved Ducts

Download Full-text

Aerodynamic Inverse Design of Transonic Compressor Cascades with Stabilizing Elastic Surface Algorithm

Applied Sciences ◽

10.3390/app11114845 ◽

2021 ◽

Vol 11 (11) ◽

pp. 4845

Author(s):

Mohammad Hossein Noorsalehi ◽

Mahdi Nili-Ahmadabadi ◽

Seyed Hossein Nasrazadani ◽

Kyung Chun Kim

Keyword(s):

Design Method ◽

Inverse Design ◽

Normal Shock ◽

Compressor Cascade ◽

Elastic Surface ◽

Transonic Compressor ◽

Pressure Distributions ◽

The Difference ◽

Inverse Design Method ◽

Transonic Cascade

The upgraded elastic surface algorithm (UESA) is a physical inverse design method that was recently developed for a compressor cascade with double-circular-arc blades. In this method, the blade walls are modeled as elastic Timoshenko beams that smoothly deform because of the difference between the target and current pressure distributions. Nevertheless, the UESA is completely unstable for a compressor cascade with an intense normal shock, which causes a divergence due to the high pressure difference near the shock and the displacement of shock during the geometry corrections. In this study, the UESA was stabilized for the inverse design of a compressor cascade with normal shock, with no geometrical filtration. In the new version of this method, a distribution for the elastic modulus along the Timoshenko beam was chosen to increase its stiffness near the normal shock and to control the high deformations and oscillations in this region. Furthermore, to prevent surface oscillations, nodes need to be constrained to move perpendicularly to the chord line. With these modifications, the instability and oscillation were removed through the shape modification process. Two design cases were examined to evaluate the method for a transonic cascade with normal shock. The method was also capable of finding a physical pressure distribution that was nearest to the target one.

Download Full-text

Study on the choices of design parameters for inverse design of metasurface using Deep leargning

2020 IEEE MTT-S International Conference on Numerical Electromagnetic and Multiphysics Modeling and Optimization (NEMO) ◽

10.1109/nemo49486.2020.9343384 ◽

2020 ◽

Author(s):

Junjie Hou ◽

Hai Lin ◽

Lijie Chen ◽

Feng Deng ◽

Zhonghua Fang

Keyword(s):

Inverse Design ◽

Design Parameters

Download Full-text