Long Time Asymptotic Behavior for the Nonlocal mKdV Equation in Solitonic Space–Time Regions

We study the long time asymptotic behavior for the Cauchy problem of an integrable real nonlocal mKdV equation with nonzero initial data in the solitonic regions

$$\begin{aligned}&q_t(x,t)-6\sigma q(x,t)q(-x,-t)q_{x}(x,t)+q_{xxx}(x,t)=0, \\&\quad q(x,0)=q_{0}(x),\ \ \lim _{x\rightarrow \pm \infty } q_{0}(x)=q_{\pm }, \end{aligned}$$

where \(|q_{\pm }|=1\) and \(q_{+}=\delta q_{-}\), \(\sigma \delta =-1\). In our previous article, we have obtained long time asymptotics for the nonlocal mKdV equation in the solitonic region \(-6<\xi <6\) with \(\xi =\frac{x}{t}\). In this paper, we give the asymptotic expansion of the solution q(x, t) for other solitonic regions \(\xi <-6\) and \(\xi >6\). Based on the Riemann–Hilbert formulation of the Cauchy problem, further using the \({\bar{\partial }}\) steepest descent method, we derive different long time asymptotic expansions of the solution q(x, t) in above two different space-time solitonic regions. In the region \(\xi <-6\), phase function \(\theta (z)\) has four stationary phase points on the \({\mathbb {R}}\). Correspondingly, q(x, t) can be characterized with an \({\mathcal {N}}(\Lambda )\)-soliton on discrete spectrum, the leading order term on continuous spectrum and an residual error term, which are affected by a function \(\textrm{Im}\nu (\zeta _i)\). In the region \(\xi >6\), phase function \(\theta (z)\) has four stationary phase points on \(i{\mathbb {R}}\), the corresponding asymptotic approximations can be characterized with an \({\mathcal {N}}(\Lambda )\)-soliton with diverse residual error order \({\mathcal {O}}(t^{-1})\).