2016 ANNUAL REVIEW Ministry of Science and Technology

The two foregoing research results represented pioneering efforts in both obtaining valid solutions and also performing in-depth theoretical exploration of extreme high dimensional semiparametric model variable selection and model structure determination problems. (3) World-leading advances in near-infrared light- emitting materials and elements A research team made up of researchers from National Taiwan University and National Tsinghua University has overcome theoretical restrictions to design and synthesize a platinum complex compound that can self-assemble into a linear arrangement. The material's perfect molecular arrangement extends its emission wavelength to the near infrared region (740 nm), and, from a theoretical perspective, weakens the interaction of excitons and phonons, which strengthens exciton emissions. Using advanced process technology, the research team set a new world record by boosting the efficiency of near infrared organic LEDs by a factor of nearly ten, and achieved an external quantum efficiency allowing commercial production (24%). With regard to near infrared organic LEDs (OLEDs), because the human eye cannot see near infrared light, this technology cannot be used to produce display elements and lighting, it can be used in other types of applications, and is expected to provide new opportunities to domestic industry. The most promising potential application consists of near infrared illumination units for vehicle collision sensors. Because near infrared OLEDs offer such advantages as planar emissions, diversity of product type, and inexpensive production, they may be able to replace the LEDs currently used in vehicle sensors, and also found all important explanatory variables, and determined whether they had a constant coefficient of variation or nonconstant coefficient of variation. The method's two steps consisted of the selection of variables and determination of model structure, as follows: In the first step, the "sure independence screening" variable screening method was used to construct a marginal varying-coefficient model for each explanatory variables, and the L2 norms of the estimated values of the variation coefficient function were ranked from large to small. Afterwards, most of the least important explanatory variables were screened out. In the second step, group SCAD was used to form the group SCAD penalty term of the L2 norm of the fit spline variation coefficient function and deviation constant of the explanatory variables remaining from screening in the first step. This process not only obtained the important explanatory variables, but also automatically classified them on the basis of whether they had a constant coefficient of variation or nonconstant coefficient of variation. In 2016, the research team also proposed a variable screening method for use in extreme high dimensional coefficient of variation regression models in a paper published in the Journal of American Statistical Association . This paper employed stepwise procedures to perform the variable screening process. In this stepwise variable screening method, the residual sum of squares was used as a standard for further selection of potential variables, and BIC was used as a guideline concerning whether to stop variable selection. This method can also achieve variable selection consistency under some conditions, which involves discarding explanatory variables with parametric functions that are zero functions and retaining explanatory variables with parametric functions that are non-zero functions. Synthesis of a platinum complex compound that can self-assemble into a linear arrangement. The material's perfect molecular arrangement extends its emission wavelength to the near infrared region (740 nm), and, from a theoretical perspective, weakens the interaction of excitons and phonons, which strengthens exciton emissions. Self-assembly of molecules into a linear arrangement in a solid or film Individual molecules in solution Single molecule MOST Ministry of Science and Technology Ministry of Science and Technology 30 2016 ANNUAL REVIEW