Quasi-Monte Carlo convergence conjecture for mapped (0,2)(0,2)-sequences on the sphere

About 15 years old · traced to

Let x0,x1,…∈[0,1)2\boldsymbol{x}_0,\boldsymbol{x}_1,\ldots \in [0,1)^2 be a (0,2)(0,2)-sequence, and let z0,z1,…∈S2\boldsymbol{z}_0,\boldsymbol{z}_1,\ldots \in \mathbb{S}^2 be the corresponding points on the sphere obtained using the mapping Φ\Phi. For N≥1N\geq 1, define the quadrature rule

QN(f)=1N∑n=0N−1f(zn).Q_N(f)=\frac{1}{N}\sum_{n=0}^{N-1}f(\boldsymbol{z}_n).

Convergence conjecture. The squared worst-case error in the Sobolev space H3/2\mathcal{H}^{3/2} satisfies

e2(QN,H3/2)=O(N−3/2)as N→∞.e^2(Q_N,\mathcal{H}^{3/2})=\mathcal{O}(N^{-3/2})\qquad\text{as }N\to\infty.

This would establish the expected quasi-Monte Carlo convergence rate for digital (0,2)(0,2)-sequences mapped from the unit square to S2\mathbb{S}^2. The claim is motivated by the numerical results presented in the paper; no proof or resolution is supplied in the source.

References

Primary source

Johann S. Brauchart and Josef Dick, “Quasi-Monte Carlo rules for numerical integration over the unit sphere S^2”, arXiv:1101.5450 (2011).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.