令人头疼的科研工作

我们引用Tass的文章：

• Tass, P. A., et al. “Long-lasting desynchronization in rat hippocampal slice induced by coordinated reset stimulation.” Physical Review E 80.1 (2009): 011902.
• Hauptmann, Christian, and Peter A. Tass. “Restoration of segregated, physiological neuronal connectivity by desynchronizing stimulation.” Journal of neural engineering7.5 (2010): 056008.

Tam, H. C., Emily SC Ching, and Pik-Yin Lai. “Reconstructing networks from dynamics with correlated noise.” Physica A: Statistical Mechanics and its Applications 502 (2018): 106-122.

The time evolution of the node variables is given by a set of coupled differential equations:
$$\frac{d{x}_i}{dt}=f_i(x_i)+\sum _{j̸=i}{A}_{ij}h(x_i,x_j)+{\eta }_i$$

External influences on the system are modelled by a Gaussian correlated noise ${\eta }_i$ that is generated by the Ornstein-Uhlenbeck process:
$${\tau }_n\frac{d{\eta }_i}{dt}=-{\eta }_i+{\xi }_i$$

where ${\xi }_i(t)$ is a zero-mean Gaussian white noise:
$$<{\xi }_i(t)>=0,<{\xi }_i(t){\xi }_j(t’)>=2D_{ij}\delta (t-t’)$$

Thus we have
$$<{\eta }_i(t)>=0,<{\eta }_i(t){\eta }_j(t’)>=\frac{D_{ij}}{{\tau }_n}{e}^{-|t-t’|/{\tau }_n}$$

线性化可得：
$$\frac{d}{dt}\delta x=Q\delta x+{\eta }_i$$

对上式积分可得：
$$\delta x(t)=e^{Qt}\delta x(0)+{\int }_0^t{e}^{(t-t’)Q}{\eta }_i(t’)dt’$$

Hence we obtain
$$K_{\tau }=<\delta x(t+\tau )\delta x(t{)}^T>$$

$$Q{K}_0+K_0{Q}^T+D(I-{\tau }_n{Q}^T{)}^{-1}+(I-{\tau }_{n}Q{)}^{-1}{D}^T=0$$

$$K_{\tau }=e^{\tau Q}{K}_0+J(\tau )$$

$$J(\tau )=(e^{\tau Q}-e^{-\tau /{\tau }_n}I)U+\tau e^{\tau Q}V$$

$$U={\tau }_{n}RD(I-{\tau }_n{Q}^T{)}^{-1}$$

$$V=[I-(I+{\tau }_{n}Q)R]RD(I-{\tau }_n{Q}^T{)}^{-1}$$

$$R=(I+{\tau }_{n}Q{)}^{-1}$$

Replacing $\tau$ by $2\tau$ and then by $3\tau$ and after some algebra, we obtain
$$K_{2\tau }=S{K}_{\tau }-T{K}_0$$

$$K_{3\tau }=S{K}_{2\tau }-T{K}_{\tau }$$

$$S=e^{Q\tau }+e^{-\tau /{\tau }_n}I$$

$$T=e^{-\tau /{\tau }_n}{e}^{Q\tau }$$

Solving Eqs, we then obtain
$$T=(K_{3\tau }-K_{2\tau }{K}_{\tau }^{-1}{K}_{2\tau })(K_0{K}_{\tau }^{-1}{K}_{2\tau }-K_{\tau }{)}^{-1}$$

$$S=(K_{2\tau }+T{K}_{\tau })K_0^{-1}$$

$$S=e^{\tau /{\tau }_n}T+e^{-\tau /{\tau }_n}I$$

to estimate the value of $e^{\tau /{\tau }_n}$ and thus ${\tau }_n$ by a least-square fit of the diagonal elements of $S$, denoted by $\mu$,

$$Q=\frac{1}{\tau }log(\mu T)=\frac{1}{\tau }[log(\mu )+logT]$$