Table of Contents
Fetching ...

Comment and correction for "On Explicit Construction of Simplex $t$-designs" by M. S. Baladram

Jakub Czartowski

TL;DR

This work identifies and corrects an error in Baladram's claim of a 3-point simplex $3$-design in the $d$-point probability simplex by showing the cyclic symmetry-based argument is invalid. It introduces a corrected framework using the full symmetric group $\mathcal{S}_d$ to restore moment-equality conditions and further proposes symmetry-restricted simplex designs as a robust alternative. The authors extend the construction to a $d(d-1)$-point family that yields simplex $3$-designs for small dimensions and conjecture a general $d^{t-1}$ scaling for minimal designs, offering concrete equations and feasibility analyses. These contributions clarify the proper group-theoretic approach to simplex designs and open avenues for scalable, symmetry-aware design constructions with potential applications in numerical integration and related fields.

Abstract

In [Bal18] a new method of constructing simplex designs based on cyclic group on $n$ elements has been proposed. One of the claims put forward therein is existence of 3-point simplex 3-design in dimension $d = 3$. In this manuscript we present explicit counterarguments and suggest a manner to rectify the existing proofs. By doing this, we show that the results presented in [Bal18] can be utilised to construct simplex 3-designs scaling as $d^2$, which suggest a general scaling of $d^{t-1}$. Finally, we put forward a notion that encompasses the objects conforming with bounds given in [Bal18], which we refer to as symmetry-restricted simplex $t$-designs.

Comment and correction for "On Explicit Construction of Simplex $t$-designs" by M. S. Baladram

TL;DR

This work identifies and corrects an error in Baladram's claim of a 3-point simplex -design in the -point probability simplex by showing the cyclic symmetry-based argument is invalid. It introduces a corrected framework using the full symmetric group to restore moment-equality conditions and further proposes symmetry-restricted simplex designs as a robust alternative. The authors extend the construction to a -point family that yields simplex -designs for small dimensions and conjecture a general scaling for minimal designs, offering concrete equations and feasibility analyses. These contributions clarify the proper group-theoretic approach to simplex designs and open avenues for scalable, symmetry-aware design constructions with potential applications in numerical integration and related fields.

Abstract

In [Bal18] a new method of constructing simplex designs based on cyclic group on elements has been proposed. One of the claims put forward therein is existence of 3-point simplex 3-design in dimension . In this manuscript we present explicit counterarguments and suggest a manner to rectify the existing proofs. By doing this, we show that the results presented in [Bal18] can be utilised to construct simplex 3-designs scaling as , which suggest a general scaling of . Finally, we put forward a notion that encompasses the objects conforming with bounds given in [Bal18], which we refer to as symmetry-restricted simplex -designs.
Paper Structure (12 sections, 5 theorems, 35 equations, 1 figure, 2 tables)

This paper contains 12 sections, 5 theorems, 35 equations, 1 figure, 2 tables.

Key Result

Lemma 1

Let $n, k\in\mathbb{Z}_{>0},\,r_1,\,r_2\in\mathbb{R}$. Define Let $V_k$ be a subspace spanned by $F(j,0,\hdots,0), 0\leq j\leq k$. Then, for non-negative integers $k_1,\hdots,k_n$ such that $\sum_{i=1}^n k_i = k$ one has $F(k_1,\hdots,k_n)\in V_k$.

Figures (1)

  • Figure 1: Monomials and symmetrised monomials in 3-point simplex: In the following figures we show contour plots of monomials $M(\vb{k})$ and symmetrised monomials $F_G(\vb{k})$ in 3-points simplex for $\sum k_i \leq 3$ and $G =\mathcal{C}_3,\,\mathcal{S}_3$; Additionally, we show cyclically symmetrised solution of \ref{['eq:3-point_non_des']} (red) and its mirror image (green). Note that for all $F_{\mathcal{S}_3}$ all six points lie on the same isoline of the function, thus noting that average over red and green points will be equal. However, in particular case of $F_{\mathcal{C}_3}(2,1,0)$, we find that red points lie on a different isoline than green points, thus leading to different values of averages depending on the mirror symmetry -- equivalent to statement given in Eq. \ref{['eq:av_inequal']}; Additionally, $F_{\mathcal{C}_3}(2,1,0)$ is the only polynomial without mirror symmetry, and thus cannot be decomposed of the remaining functions.

Theorems & Definitions (8)

  • Lemma 1: Lemma 3.1 from baladram2018explicit with $\mathcal{S}_d$
  • Theorem 1: Theorem 3.2 from baladram2018explicit with $\mathcal{S}_d$
  • proof
  • Theorem 2
  • Conjecture 1
  • Lemma 2: Lemma 3.1 from baladram2018explicit with $G$-invariant sequences $\vb{k}$
  • Theorem 3: Theorem 3.2 from baladram2018explicit with $G$-restricted designs
  • proof