mbend: Data, Code & Results

The data, code and results from a study on R package mbend are provided. R package mbend bends symmetric non-positive-definite matrices to symmetric positive definite matrices. A 5×5 non-positive-definite covariance matrix, its conversion to a correlation matrix, a 5×5 weight matrix for weighted bending, and an ill-conditioned genomic relationship matrix (created by method 1 of VanRaden, 2008) of 5,000 SNP and 1,000 animals with 10 duplicated genotypes and no quality control were used as data. Weighted and unweighted bending methods of Jorjani et al. (2003) and Schaeffer (2014) were tested and compared with the method of Bock et al. (1988) that is implemented in R package psych.

本研究提供了针对R包mbend的相关数据、代码与实验结果。R包mbend可将对称非正定矩阵转换为对称正定矩阵。本次研究采用的测试数据集包括：一个5×5的非正定协方差矩阵、其转换得到的相关矩阵、用于加权弯折的5×5权重矩阵，以及一个由5000个单核苷酸多态性（Single Nucleotide Polymorphism, SNP）标记、1000个个体构建的病态基因组关系矩阵——该矩阵采用VanRaden于2008年提出的方法1构建，包含10个重复基因型且未进行质量控制。本研究对Jorjani等人（2003）与Schaeffer（2014）提出的加权及非加权弯折方法进行了测试，并与R包psych中实现的Bock等人（1988）提出的矩阵弯折方法进行了对比。