# Ground state for non-linear Schrödinger

When studying the blow-up for focusing non-linear Schrödinger equation (NLS) one often compares the initial-state to a stationary solution.

In the energy-critical case, this stationary solution is for example

$$\Delta W+ \vert W\vert^{\frac{4}{N-2}}W=0$$

see Kenig-Merle.

For the mass-critical case however this ground state satisfies

$$\Delta W + \vert W\vert^{1+\frac{4}{N}}W =W.$$

What I would like to know. Why is there this additional term $$W$$ in the mass-critical case?

A useful perspective on this is given in

Weinstein, Michael I., Nonlinear Schrödinger equations and sharp interpolation estimates, Commun. Math. Phys. 87, 567-576 (1983). ZBL0527.35023.

As observed in that paper, the ground state for NLS with energy-critical or sub-critical exponent $$2 \leq p \leq 1 + \frac{4}{N-2}$$ can be viewed as the extremiser for the Gagliardo-Nirenberg inequality

$$\frac{\int_{{\bf R}^N} |u|^{p+1}}{(\int_{{\bf R}^N} |u|^2)^{1-\frac{(N-2)(p-1)}{4}} (\int_{{\bf R}^N} |\nabla u|^2)^{N(p-1)/4}}.$$

Standard Euler-Lagrange calculations then shows in the energy-subcritical case $$p < 1 + \frac{4}{N-2}$$ that these extremisers (formally at least) obey the ground state equation $$\Delta u + \alpha |u|^{p-1} u = \beta u$$ with some Lagrange multipliers $$\alpha,\beta$$ that can be renormalised to $$1$$ by suitable rescaling. But in the energy critical case $$p = 1 + \frac{4}{N-2}$$, the mass term $$\int_{{\bf R}^N} |u|^2$$ in the above functional disappears, and so does the Lagrange multiplier $$\beta$$.

In a physics context the nonlinear Schrödinger equation describes the ground state wave function $$W(\mathbf{x})$$ of a two-dimensional interacting atomic gas, with interaction potential $$m(\mathbf{x})|W|^2$$ in an external potential $$V(\mathbf{x})$$ at chemical potential $$\mu$$:
$$-\Delta W(\mathbf{x})+V(\mathbf{x})W(\mathbf{x})+\mu W(\mathbf{x})=m(\mathbf{x})|W(\mathbf{x})|^2 W(\mathbf{x}).$$
For the mass-critical case one has $$m(\mathbf{x})=\pm 1$$. The equation in the OP corresponds to the case $$m=1$$, $$V=0$$, $$\mu=1$$, so a mass-critical system at nonzero chemical potential. As explained in Threshold behavior and uniqueness of ground states for mass critical inhomogeneous Schrödinger equation, the solution for $$\mu=1$$ can be mapped onto the solution for other values of $$\mu$$.
• so why do we now have the choices $\mu=1$ and $\mu=0$ in the two scenarios? – Sascha Aug 28 '19 at 10:47
• $\mu\neq 0$ spoils the energy-criticality; for the mass-critical case you only want the mass term $m$ to be unaffected by the scaling, so you can allow for $\mu\neq 0$. – Carlo Beenakker Aug 28 '19 at 11:45