2025年度第1回計算科学フォーラム

上野和雅

(東京大学大学院 理学系研究科 博士課程／株式会社QunaSys リサーチインターン)

「気象・気候分野における量子計算活用の可能性」

気象・気候分野では、数値モデルの高解像度化や大規模アンサンブル化に伴い、古典計算機による計算コストの増大が深刻な課題となっている。量子計算は、量子力学の性質や高い状態表現能力を活かすことで、特定の問題において古典計算を凌駕できる可能性がある一方、気象・気候問題への適用指針は未整備である。本講演では、雲粒分布・流体・定常放射伝達を対象とした量子アルゴリズム設計の取り組みを紹介する。さらに、富岳を用いたシミュレーションによる検証結果と量子計算リソースの見積もり結果を示し、実用化に向けた展望を議論する。

Enhua Xu

（理化学研究所 計算科学研究センター 量子系分子科学研究チーム 研究員）

「A scalable diagonalization framework for tensor-product bitstring selected configuration interaction」

Selected configuration interaction (SCI) methods are effective for treating strongly correlated electronic systems, yet their scalability has long been limited by implementations that replicate the configuration interaction (CI) vector across processes, resulting in severe memory bottlenecks.
Here, we present a fully distributed SCI framework grounded in a tensor-product bitstring (TPB) representation, in which determinants are organized through a TPB structure constructed from selected alpha- and beta-bitstrings. We refer to this approach as tensor-product bitstring SCI (TBSCI).
An efficient TBSCI eigensolver is developed together with a bitstring-based Hamiltonian evaluation algorithm that enables on-the-fly Hamiltonian application with computational cost approximately proportional to the number of retained determinants.
In addition, a suite of MPI communication optimization strategies is integrated to sustain parallel efficiency at large node counts.
Large-scale full configuration interaction (FCI) benchmarks, used as communication-intensive stress tests, demonstrate that the implemented TBSCI eigensolver continues to reduce wall time for distributed diagonalization of 2.6 trillion determinants, reaching 54,000 nodes (more than 2.5 million cores) on supercomputer Fugaku.
Beyond scalability, we investigate the structural compactness of the TPB representation and show that selecting alpha- and beta-bitstrings according to their collective weights in a reference SCI wavefunction yields TPB-based wavefunctions that approach the FCI limit while using only a small fraction of determinants in the tested systems.
These results establish TBSCI as a scalable diagonalization framework and provide evidence for the structural compactness of the TPB representation.