• ### Unified phase diagram for iron-based superconductors(1704.05182)

Oct. 17, 2017 cond-mat.supr-con
High-temperature superconductivity is closely adjacent to a long-range antiferromagnet, which is called a parent compound. In cuprates, all parent compounds are alike and carrier doping leads to superconductivity, so a unified phase diagram can be drawn. However, the properties of parent compounds for iron-based superconductors show significant diversity and both carrier and isovalent dopings can cause superconductivity, which casts doubt on the idea that there exists a unified phase diagram for them. Here we show that the ordered moments in a variety of iron pnictides are inversely proportional to the effective Curie constants of their nematic susceptibility. This unexpected scaling behavior suggests that the magnetic ground states of iron pnictides can be achieved by tuning the strength of nematic fluctuations. Therefore, a unified phase diagram can be established where superconductivity emerges from a hypothetical parent compound with a large ordered moment but weak nematic fluctuations, which suggests that iron-based superconductors are strongly correlated electron systems.
• ### Adaptive Elastic Net Method for Cox Model(1507.06371)

July 23, 2015 math.PR, math.ST, stat.TH
In this paper, we study the Adaptive Elastic Net method for the Cox model. We prove the grouping effect and oracle property of its estimators. Finally, we show these two properties by an empirical analysis and a numerical simulation, respectively.
• ### Elastic Net Procedure for Partially Linear Models(1507.06032)

July 22, 2015 math.PR, stat.ME, stat.ML
Variable selection plays an important role in the high-dimensional data analysis. However the high-dimensional data often induces the strongly correlated variables problem. In this paper, we propose Elastic Net procedure for partially linear models and prove the group effect of its estimate. By a simulation study, we show that the strongly correlated variables problem can be better handled by the Elastic Net procedure than Lasso, ALasso and Ridge. Based on an empirical analysis, we can get that the Elastic Net procedure is particularly useful when the number of predictors $p$ is much bigger than the sample size $n$.