Nonuniform Bose-Einstein condensate. I. An improvement of the Gross-Pitaevskii method
A nonuniform condensate is usually described by the Gross-Pitaevskii (GP) equation, which is derived with the help of the c-number ansatz $\hat{ Ψ}(\mathbf{r},t)=Ψ(\mathbf{r},t)$. Proceeding from a more accurate operator ansatz $\hatΨ(\mathbf{r},t)=\hat{a}_{0}Ψ(\mathbf{r},t)/ \sqrt{N}$, where $N$ is the number of Bose particles, we find the equation $i\hbar \frac{\partial Ψ(\mathbf{r},t)}{\partial t}=-\frac{\hbar ^{2}}{2m}\frac{\partial ^{2}Ψ(\mathbf{r},t)}{\partial \mathbf{r}^{2}}+\left( 1-\frac{1}{N}\right) 2cΨ(\mathbf{r},t)|Ψ(\mathbf{r},t)|^{2}$, which we call the GP$_{N}$ equation. It differs from the GP equation by the factor $\left( 1-1/N\right) $. We compare the accuracy of the GP and GP$_{N}$ equations by analysing the ground state of a one-dimensional system of point bosons with repulsive interaction ($c>0$) and zero boundary conditions. Both equations are solved numerically, and the system energy $E$ and the particle density profile $ρ(x)$ are determined for the mean particle density $\barρ=1$, different values of $N$ and of the coupling constant $γ=c/\barρ$. The solutions are compared with the exact ones obtained by the Bethe ansatz. The results show that the GP and GP$_{N}$ equations equally well describe the many-particle system ($N\gtrsim 100$) being in the weak coupling regime ($N^{-2}\ll γ\lesssim 0.1$). But for the few-boson system ($N\lesssim 10$) with $γ\lesssim N^{-2}$ the solutions of the GP$_{N}$ equation are in much better agreement with the exact ones. Thus, the multiplier $\left( 1-1/N \right) $ allows one to describe few-boson systems with high accuracy. This means that it is reasonable to extend the notion of Bose-Einstein condensation to few-particle systems.