【学术】重构具有时间延迟相互作用的动力学网络

Reconstruction of dynamic networks with time-delayed interactions in presence of fast-varying noises

Zhaoyang Zhang	Yang Chen	Yuanyuan Mi	Gang Hu
Ningbo University	中科院脑网中心和国家模式识别实验室	Chongqing University	Beijing Normal University

Dated: April 1, 2019

理论推导

考虑$N$个节点的动力学系统
$${\dot{x}}_i=F_i[x_i(t)]+\sum _{j=1,j̸=i}^N{\Phi }_{ij}[x_i(t),x_j(t-{\tau }_{ij})]+{\eta }_i(t)+{\Gamma }_i(t),i=1,2,...,N$$

其中${\eta }_i(t)$为色噪声，${\Gamma }_i(t)$为白噪声，满足
$$<{\eta }_i(t)>=0,<{\eta }_i(t){\eta }_i(t+t^{\prime })>=P_{ij}{e}^{-\frac{|t^{\prime }|}{{\tau }_C}}$$

$$<{\Gamma }_i(t)>=0,<{\Gamma }_i(t){\Gamma }_i(t+t^{\prime })>=Q_i{\delta }_{ij}\delta (t^{\prime })$$

$$<{\Gamma }_i(t){\eta }_j(t)>=0$$

$F_i$和${\Phi }_{ij}$可以为线性或者非线性。需要设计算法，从数据中重构出网络连接。假设在整个网络中我们只能够测量两个节点$A$和$B$，并且有充足的数据。
$$x_A(t)=[x_A(t_1),x_A(t_2)....,x_A(t_k),...,x_A(t_L)]$$

$$x_B(t)=[x_B{t}_1),x_B(t_2)....,x_B(t_k),...,x_B(t_L)]$$

$$0<\Delta t=t_{k+1}-t_k\ll 1,L\gg 1$$

对$x_i(t)$求时间的二阶段导可得
$${\ddot{x}}_i(t)=\frac{\partial F_i[x_i(t)]}{\partial x_i(t)}{\dot{x}}_i(t)+\sum _{j=1,j̸=i}^N\frac{\partial {\Phi }_{ij}[x_i(t),x_j(t-{\tau }_{ij})]}{\partial x_i(t)}{\dot{x}}_i(t)+\sum _{j=1,j̸=i}^N\frac{\partial {\Phi }_{ij}[x_i(t),x_j(t-{\tau }_{ij})]}{\partial x_j(t-{\tau }_{ij})}{\dot{x}}_j(t-{\tau }_{ij})+{\dot{\eta }}_i(t)+{\dot{\Gamma }}_i(t)$$

其中高阶导数可以使用前向差分进行计算
$${\dot{x}}_i(t_k)=\frac{x_i(t_{k+1})-x_i(t_k)}{\Delta t},{\ddot{x}}_i(t_k)=\frac{{\dot{x}}_i(t_{k+1})-{\dot{x}}_i(t_k)}{\Delta t}$$

噪声的导数被定义为
$${\dot{\eta }}_i(t_k)=\frac{{\eta }_i(t_{k+1})-{\eta }_i(t_k)}{\Delta t},{\dot{\Gamma }}_i(t_k)=\frac{{\Gamma }_i(t_{k+1})-{\Gamma }_i(t_k)}{\Delta t}$$

根据前面的公式可知
$${\ddot{x}}_A(t_k)=\frac{1}{2}\frac{\partial F_A[x_A(t_k)]}{\partial x_i(t_k)}[{\dot{x}}_A(t_k)+{\dot{x}}_A(t_{k+1})]+\frac{1}{2}\sum _{j=1,j̸=A}^N\frac{\partial {\Phi }_{Aj}[x_A(t),x_j(t_k-{\tau }_{Aj})]}{\partial x_A(t_k)}[{\dot{x}}_A(t_k)+{\dot{x}}_A(t_{k+1})]+\frac{1}{2}\sum _{j=1,j̸=A}^N\frac{\partial {\Phi }_{Aj}[x_i(t),x_j(t-{\tau }_{Aj})]}{\partial x_j(t-{\tau }_{Aj})}[{\dot{x}}_j(t_k-{\tau }_{Aj})+{\dot{x}}_j(t_{k+1})]+\frac{{\eta }_A(t_{k+1-{\tau }_{Aj}})-{\eta }_A(t_k)}{\Delta t}+\frac{{\Gamma }_A(t_{k+1})-{\Gamma }_A(t_k)}{\Delta t}$$

然后对方程左右两边同乘$x_B(t_k+\Delta t)$，计算每项的关联可得
$$R_{AB}=<{\ddot{x}}_A(t_k)x_B(t_k+\Delta t)>$$

$$=<\frac{1}{2}\frac{\partial F_A[x_A(t_k)]}{\partial x_i(t_k)}[{\dot{x}}_A(t_k)x_B(t_k+\Delta t)+{\dot{x}}_A(t_{k+1}{x}_B(t_k+\Delta t))]>$$

$$+\frac{1}{2}\sum _{j=1,j̸=A}^N\frac{\partial {\Phi }_{Aj}[x_A(t),x_j(t_k-{\tau }_{Aj})]}{\partial x_A(t_k)}[{\dot{x}}_A(t_k)x_B(t_k+\Delta t)+{\dot{x}}_A(t_{k+1})x_B(t_k+\Delta t)]$$

$$+\frac{1}{2}\sum _{j=1,j̸=A}^N\frac{\partial {\Phi }_{Aj}[x_i(t),x_j(t-{\tau }_{Aj})]}{\partial x_j(t-{\tau }_{Aj})}[{\dot{x}}_j(t_k-{\tau }_{Aj})x_B(t_k+\Delta t))+{\dot{x}}_j(t_{k+1})x_B(t_k+\Delta t))]$$

$$+\frac{{\eta }_A(t_{k+1-{\tau }_{Aj}})x_B(t_k+\Delta t)-{\eta }_A(t_k)x_B(t_k+\Delta t)}{\Delta t}+\frac{{\Gamma }_A(t_{k+1})x_B(t_k+\Delta t)-{\Gamma }_A(t_k)x_B(t_k+\Delta t)}{\Delta t}$$

记该公式为方程1。因为有白噪声的存在，对$<{\dot{x}}_i(t){\dot{x}}_j(t+t^{\prime })>$积分在$t^{\prime }=0$处有个阶跃，所以
$$<{\dot{x}}_i(t)x_j(t+\Delta t)>-<{\dot{x}}_i(t)x_j(t)>=Q_j{\delta }_{ij}$$

定义$n_{AB}=\frac{{\tau }_{AB}}{\Delta t}$，$V_i(\Delta k)=<{\dot{x}}_i(t_k)x_i(t_{k+\Delta k})>$，在$\Delta k=0$到$\Delta k=1$处有
$$V_i(1)-V_i(0)=Q_i$$

由于$<{\dot{x}}_B(t_k-{\tau }_{AB})x_B(t_{k+\Delta k})>$和$<{\dot{x}}_B(t_{k+1}-{\tau }_{AB})x_B(t_{k+\Delta k})>$在$k=-n_{AB}$处不联系，可以得到$n_{AB}$即${\tau }_{AB}$
根据$<{\dot{x}}_i(t)x_j(t+\Delta t)>-<{\dot{x}}_i(t)x_j(t)>=Q_j{\delta }_{ij}$的性质可知方程1右边除了$<{\dot{x}}_B(t_k-{\tau }_{AB})x_B(t_{k+\Delta k})>$和$<{\dot{x}}_B(t_{k+1}-{\tau }_{AB})x_B(t_{k+\Delta k})>$没有不连续的项，所以从方程1可以得到
$$D_{AB}=R_{AB}(-n_{AB}+2)-R_{AB}(-n_{AB})=<{\ddot{x}}_A(t_k)x_B(t_{t_k-n_{AB}+2})>-<{\ddot{x}}_A(t_k)x_B(t_{t_k-n_{AB}})>$$

$$=<\frac{\partial {\Phi }_{Aj}[x_i(t),x_j(t-{\tau }_{Aj})]}{\partial x_B(t-{\tau }_{AB})}>Q_B$$

定义$M_{AB}=<\frac{\partial {\Phi }_{Aj}[x_i(t),x_j(t-{\tau }_{Aj})]}{\partial x_B(t-{\tau }_{AB})}>$，可得
$$M_{AB}=\frac{D_{AB}}{Q_B}=J_{AB}$$

数值仿真

定义均方根误差$E$作为衡量$M_{AB}$和$J_{AB}$之间的误差，$T$表示测量时长
对于线性系统：

对于扩散耦合的FHN网络（上），和Rossler网络（下）

对于基因调控网络

总结

文章的核心亮点是重构有时滞作用的系统，trick是利用白噪声的性质。