国债发行成本组合函数研究

由于实施积极的财政政策和适度宽松的货币政策,预计未来相当一段时期内我国国债发行规模将持续保持高位。在此背景下,研究控制国债发行成本、均衡国债发行结构具有重要现实意义。针对目前我国国债发行规模上升较快、中、长期国债发行品种集中、国债还本付息不均匀、国债发行成本较高的发行现状,本文提出一种新的国债发行成本组合函数,涵盖了制约国债发行成本的发行额、发行期限、利率期限结构以及票面利率等多种影响因素,提高了测算成本的准确度。由于采用了久期对不同债券的成本进行调整,新的国债发行成本组合函数不仅反映了债券在有效到期时间内发行成本的平均值,而且对长、短期国债的发行成本进行了平滑,在一定程度上避免了由于长期国债发行成本权重较大而造成国债发行集中在短期的现象,有助于促进国债发行结构和存量结构的均衡合理。

General Pattern and Neighborhood Search Algorithms for Non-Convex Unit Commitment

This paper presents a new approach to solve mixed-variable unit commitment(UC) problems with non-smooth cost functions based on a generalized pattern search filter(GPS-filter) algorithm. A GPS-filter algorithm does not require any information about the gradient of the objective function while searching for an optimum solution. At the same time, it is available for solving mixed-variable optimization problems, which is very suitable for UC. A new suitable discrete neighborhood structure with UC characteristics is proposed to improve GPS-filter efficiently. A lot of multiple units' states are fixed before search; hence, the polling search of discrete variable is efficient for a few uncertain units. Numerical experiments are included to demonstrate the proposed approach's ability to handle the highly nonlinear, discontinuous, non-smooth cost functions and mixed variables of the UC problem.