@techreport{oai:ipsj.ixsq.nii.ac.jp:00218766,
 author = {前蔵, 遼 and 鈴木, 泰成 and 吉岡, 信行 and 徳永, 裕己},
 issue = {7},
 month = {Jun},
 note = {近年，NISQ デバイスを用いた量子化学計算や量子機械学習などの様々な応用が研究されている．しかし，NISQ デバイスでは大量の量子ビットを必要とする量子誤り訂正を行うことができず，計算中のノイズの影響が大きいという問題がある．そこで，統計的な処理を行うことでエラーを低減でき，NISQ デバイスでも実行可能な量子エラー抑制が注目されている．本研究では，生成モデルを用いてノイズのある測定データから量子状態を再構成し，ノイズレスなオブザーバブルの期待値推定を行った．, To obtain reliable computational results with noisy quantum computers, we need to suppress computational errors stemming from high error rates of qubits. Thus, quantum error mitigation methods are eagerly explored since they can suppress physical errors without requiring additional qubits. The virtual distillation method is expected as an efficient error mitigation method. However, the applicable scope of the virtual distillation method is restricted since it requires several copies of quantum states at the same time, which is difficult to achieve with the current devices. In this paper, we propose a novel method for estimating the noiseless expectation values from several measurement results of noisy quantum states. The key of our idea is to train neural-network quantum states to reproduce the measurement distribution of the unknown quantum states, and apply virtual distillation method to the trained neural-network quantum states.},
 title = {生成モデルを用いた量子状態トモグラフィーに基づくノイズレスな期待値の推定},
 year = {2022}
}