ホットメルト型粘着剤の開発効率向上を目的として，配合設計に対してベイズ最適化に基づく適応的実験計画法を適用し，その効果について検証を行った。その結果，低粘度かつ強粘着力という従来両立が困難であった相反する目標物性に対して，いずれの目標値も満たす配合をわずか4 サイクル目の実験によって特定させた。これは，ガウス過程回帰を用いて高い予測精度を有する機械学習モデルを構築できたことで，配合空間の目的関数を的確に捉えられた可能性が高く，それによりベイズ最適化による配合空間の探索が効率化し，非常に少ない実験回数で目標物性値を得る事に成功したと考えられた。これらの結果から，ベイズ最適化に基づく実験計画法がホットメルト型粘着剤の開発において非常に有効な手段であることを確認した。

The deadline for achieving the Sustainable Development Goals（SDGs）is approaching, and realistic and immediate responses are required. In particular, concrete measures to address environmental issues such as microplastics and marine plastics are one of the most urgent issues. In the cutting-edge research field of polymer synthesis, attention is focused on how to impart degradability to conventionally used non-degradable synthetic polymers, and competitive efforts are underway around the world to design molecules and develop reactions to incorporate decomposition functions into molecules at the polymer synthesis stage. In this review, we focus on imparting degradability to vinyl polymers synthesized by radical polymerization. After explaining the basic matters related to the decomposition of vinyl polymers, we introduce degradable polymers using radical ring-opening polymerization and radical copolymerization. Research trends in controlled polymerization and depolymerization are also explained.

This article describes how computational simulations, employing density functional theory and molecular dynamics methods, can be utilized to predict material properties, thereby streamlining the research development process. By predicting properties through simulations, the number of experimental trials in material development can be reduced, leading to significant cost savings. While implementing simulations in research development, the primary challenge often lies not in computational resources or technical learning costs, but in the human operational costs. However, this can be resolved through automation using programming techniques. When applying simulations to organic material development, it is essential to strike a right balance between achieving practical results and computational costs, rather than pursuing perfect agreement with experimental data. In some cases, it is difficult to directly calculate the main properties of a material, identifying other correlated properties can be a valuable strategy.