Solutions to the nonlinear Schrödinger equation with sequences of initial data converging to a Dirac mass (Q611859)

The aim of the article is to analyze solutions of the defocussing nonlinear Schrödinger equation (NLS) with sequences of initial data that converge to a Dirac mass. One such sequence of initial data (among several that studied) leads to the initial value problem \[ i\varphi_{i}+\varphi_{xx}-2|\varphi|^{2}\varphi=0,\qquad \varphi(x,t=0)=\tfrac{1}{\varepsilon}f\big(\tfrac{x}{\varepsilon}\big), \] for a sufficiently general class of functions \(f(y)\). For convenience, it is assumed that \(\int_{\mathbb{R}}f(y)\,dy=1\) so that \(\frac{1}{\varepsilon}f(\frac{x}{\varepsilon})\) converges weakly to a unit Dirac mass \(\varepsilon\rightarrow 0.\) However, all the derived formulae are valid when \(f(y)\) has a finite non-zero mass. The authors use a combination of relatively classical results concerning the asymptotic behaviour of the integrable NLS equation together with some more involved Riemann-Hilbert analysis.