第一部門：物理数学力学問題 （A）

[S01A-04] Three-dimensional subsurface modeling using relaxed lasso

*Takayuki Shuku¹, Ikumasa Yoshida² (1. Okayama University, 2. Tokyo City University)

Keywords:subsurface modeling, sparse

This study proposes a method for three-dimensional subsurface modeling based on relaxed lasso. Sparse modeling, or least absolute shrinkage and selection operator (lasso), has received attention in geotechnical engineering because of its performance in solving under-determined inverse problems, that commonly encounter in practice. Lasso, however, can be difficult to apply for large scale problems because of its heavy computation. This study newly developed an algorithm to solve relaxed lasso problems within a reasonable time with an ordinary computer that commonly used in practice. The performance of the proposed method was investigated through some numerical tests and real case histories.