Extrapolated multilevel variance-decay conjecture

At least 1 year old · documented by

Let \prescriptExtΦℓ\prescript{\mathrm{Ext}}{}{\Phi}^{\ell} and \prescriptExtΨℓ\prescript{\mathrm{Ext}}{}{\Psi}^{\ell} denote the extrapolated fine- and coarse-level quantities at level ℓ\ell. Assume that c1,c2∈Rc_1,c_2\in\mathbb{R}. Extrapolated multilevel variance-decay conjecture. The per-level variance satisfies

Var⁡[\prescriptExtΦℓ−\prescriptExtΨℓ]≲2−2ℓmax⁡(c12−2ℓ,c22−2ℓ)=O(2−4ℓ).\operatorname{Var}\left[\prescript{\mathrm{Ext}}{}{\Phi}^{\ell}-\prescript{\mathrm{Ext}}{}{\Psi}^{\ell}\right]\lesssim 2^{-2\ell}\max\left(c_1 2^{-2\ell},c_2 2^{-2\ell}\right)=\mathcal{O}(2^{-4\ell}).

The extrapolated antithetic construction is intended to raise the time-difference contribution to second order, yielding the stated fourth-order variance decay, while leaving the computational cost per sample unchanged.

References

Primary source

Nadhir Ben Rached, Abdul-Lateef Haji-Ali, Raúl Tempone and Leon Wilkosz, “Forward Propagation of Low Discrepancy Through McKean-Vlasov Dynamics: From QMC to MLQMC”, arXiv:2409.09821 (2024).

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.