arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.01321math.PRcs.NAmath.NAmath.OC
收藏

单调系数平均场方程的最优控制及其在神经科学中的应用

Optimal control of mean field equations with monotone coefficients and applications in neuroscience

Antoine Hocquet, Alexander Vogler

摘要:

我们感兴趣的最优控制问题与某些二次成本泛函依赖于解决$X=X^α$的随机平均场型发展方程在$\mathbb R^d$ $dX_t=b(t,X_t,\mathcal L(X_t),α_t)dt+σ(t,X_t,\mathcal L(X_t),α_t)dW_t,$ $X_0\sim μ$给定的,假设下,封闭系统的FitzHugh-Nagumo神经元网络,并在实际的控制$α_t$是确定性的。为了做到这一点,我们假设,我们给出了一个漂移系数,满足单侧Lipshitz条件,并且动力学受到$π(X_t)\leq0$形式的(凸)水平集约束。我们提出的数学处理遵循最近的专着卡莫纳和Delarue类似的控制问题与Lipshitz系数的线。通过鞅方法解决极小值的存在性后,我们显示了最大值原理,然后数值研究的梯度算法的近似的最优控制。

英文摘要:

We are interested in the optimal control problem associated with certain quadratic cost functionals depending on the solution $X=X^α$ of the stochastic mean-field type evolution equation in $\mathbb R^d$ $dX_t=b(t,X_t,\mathcal L(X_t),α_t)dt+σ(t,X_t,\mathcal L(X_t),α_t)dW_t,$ $X_0\sim μ$ given, under assumptions that enclose a sytem of FitzHugh-Nagumo neuron networks, and where for practical purposes the control $α_t$ is deterministic. To do so, we assume that we are given a drift coefficient that satisfies a one-sided Lipshitz condition, and that the dynamics is subject to a (convex) level set constraint of the form $π(X_t)\leq0$. The mathematical treatment we propose follows the lines of the recent monograph of Carmona and Delarue for similar control problems with Lipshitz coefficients. After addressing the existence of minimizers via a martingale approach, we show a maximum principle and then numerically investigate a gradient algorithm for the approximation of the optimal control.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑