Dice Question Streamline Icon: https://streamlinehq.com

Resolve the discrepancy in the 2D first turning point for the cube-domain Bratu equation

Ascertain which numerical estimate is more accurate for the first turning point parameter \lambda^* of the two-dimensional Bratu equation \Delta u + \lambda e^{u} = 0 on the unit square [0,1]^2 with homogeneous Dirichlet boundary conditions—6.808124408 computed via the symmetric finite difference method or 6.808124423 reported by Doedel and Sharifi (2000)—and, thereby, determine the more precise value of \lambda^*.

Information Square Streamline Icon: https://streamlinehq.com

Background

For the 2D Bratu equation on the unit square, the authors compute a first turning point of 6.808124408 using their symmetric finite difference method, which is extremely close to the earlier value 6.808124423 reported in the literature. They note, however, that their numerical continuation has not fully converged and that the prior paper used a limited number of mesh and collocation points.

Consequently, there is an explicit uncertainty regarding which of the two values is more accurate, motivating a targeted effort to precisely determine \lambda* in 2D on the cube domain.

References

However, since our first turning point has not yet converged (see Tab.\ \ref{tab:2dtimeratio}), and the previous work used a limited number of mesh and collocation points, it remains unclear which first turning point is more accurate.

A finite difference method with symmetry properties for the high-dimensional Bratu equation (2410.12553 - Shahab et al., 16 Oct 2024) in Section Experimental Results, Subsection "1D and 2D Bratu Equation"