Gradient Convergence of Deep Learning-Based Numerical Methods for BSDEs

来源 :数学年刊B辑(英文版) | 被引量 : 0次 | 上传用户:ghmyjp
下载到本地 , 更方便阅读
声明 : 本文档内容版权归属内容提供方 , 如果您对本文有版权争议 , 可与客服联系进行内容授权或下架
论文部分内容阅读
The authors prove the gradient convergence of the deep learning-based nu-merical method for high dimensional parabolic partial differential equations and backward stochastic differential equations,which is based on time discretization of stochastic differ-ential equations(SDEs for short)and the stochastic approximation method for noncon-vex stochastic programming problem.They take the stochastic gradient decent method,quadratic loss function,and sigmoid activation function in the setting of the neural net-work.Combining classical techniques of randomized stochastic gradients,Euler scheme for SDEs,and convergence of neural networks,they obtain the O(K-1/4)rate of gradient convergence with K being the total number of iterative steps.
其他文献
Denote a semisimple Banach algebra with an identity e by A. This paper studies the Fredholm, Weyl and Browder spectral theories in a semisimple Banach algebra,
In this paper, distributed estimation of high-dimensional sparse precision matrix is pro-posed based on the debiased D-trace loss penalized lasso and the hard t
A new theoretical thermomechanical explanation of the uneven transverse temperature distribution,along the width for thin and wide hot rolled strip was proposed
利用矩阵半张量积方法研究弱双四元数矩阵方程AX=B的解,通过提出一种新的实向量表示,将弱双四元数矩阵方程问题转化为相应实矩阵方程问题.由此得到弱双四元数方程AX=B的最小
金融危机给我国经济保持高速增长带来巨大挑战,同时也为我国区域经济框架快速构建提供了契机,国务院总理温家宝在今年政府工作报告中关于大力发展区域经济的信息说明,振兴区
基于构造的辅助函数,讨论了一类带有不完全Beta积分的高阶Cauchy中值定理,给出了这类中值定理的一般形式,并讨论了该定理“中间点”的渐进性,对已有的成果做了一些推广.
扩展了传统的异质代理商模型,加入逆风者,丰富了投资者类型,形成新的资产价格内生演化机制模型.讨论了不动点的存在性,以及相对应的稳定区域.通过数值分析,对比模型(1),模型(
Let H2(D2) be the Hardy space over the bidisk D2, and let M? = [(z??(w))2] be the submodule generated by (z??(w))2, where?(w) is a function in H∞(w). The relat
Let G be a classical group over an arbitrary field F , acting on an n-dimensional vector space V = V (n, F ) over a field F . In this paper, we classify the max
基于落锤冲击试验,以含有5个变量的泡沫铝板受冲击的能量吸收规律为研究对象,运用量纲分析法描述试验系统中各参数之间的内在联系.以落锤冲击试验为基础进行仿真试验设计,利