Polynomials, Divided Differences, and Codes

S. Venkitesh

2024-12-03

Abstract:Multivariate multiplicity codes (Kopparty, Saraf, and Yekhanin, J. ACM 2014) are linear codes where the codewords are described by evaluations of multivariate polynomials (with a degree bound) and their derivatives up to a fixed order, on a suitably chosen affine point set. While good list decoding algorithms for multivariate multiplicity codes were known in some special cases of point sets by a reduction to univariate multiplicity codes, a general list decoding algorithm up to the distance of the code when the point set is an arbitrary finite grid, was obtained only recently (Bhandari et al., IEEE TIT 2023). This required the characteristic of the field to be zero or larger than the degree bound, and this requirement is somewhat necessary, since list decoding this code up to distance with small output list size is not possible when the characteristic is significantly smaller than the degree. In this work, we present an alternate construction, based on divided differences, that closely resembles the classical multiplicity codes but is `insensitive to the field characteristic'. We obtain an efficient algorithm that list decodes this code up to distance, for arbitrary finite grids and over all finite fields. Notably, our construction can be interpreted as a `folded Reed-Muller code', which might be of independent interest. The upshot of our result is that a good `Taylor-like expansion' can be expressed in terms of a good `derivative-like operator' (a divided difference), and this implies that the corresponding code admits good algorithmic list decoding.

Information Theory,Combinatorics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: on any finite field, for multivariate multiplicity codes, design an efficient list - decoding algorithm that enables decoding when the code distance is reached. Specifically, the paper hopes to construct a new coding method based on piecewise divided differences. This coding method does not depend on the characteristics of the finite field and can perform list - decoding efficiently on any finite grid.

### Main contributions of the paper

1. **Proposed a new construction method**:
- A construction method based on piecewise divided differences. This method is very similar to the classical multiplicity codes but is not sensitive to the characteristics of the finite field.
- This construction can be interpreted as a folded Reed - Muller code, which may have independent research value.

2. **Efficient list - decoding algorithm**:
- Proposed an efficient algorithm that can perform list - decoding on this newly constructed code on any finite grid and all finite fields until the code distance.
- The concept of this algorithm is simpler than the existing decoding algorithms for classical multivariate multiplicity codes.

3. **Solved the characteristic - sensitive problem**:
- In the existing multivariate multiplicity codes, when the characteristic of the finite field is significantly smaller than the degree of the polynomial, list - decoding with a small output list size cannot be achieved. The construction method proposed in this paper overcomes this limitation.

### Formula representation

To ensure the correctness and readability of the formulas, the following are some key formulas involved in the paper:

- **Taylor expansion**:
\[
f(X)=\sum_{\alpha\in\mathbb{N}^m}\frac{D_Q^\alpha f(Y)}{[\alpha]_Q!}(X - Y)_Q^{(\alpha)}
\]
where $[\alpha]_Q! = [\alpha_1]_Q!\cdots[\alpha_m]_Q!$, and $[t]_Q=\frac{Q^t - 1}{Q - 1}$.

- **Q - derivative**:
\[
D_Q f(X)=\frac{f(QX)-f(X)}{(Q - 1)X}
\]

- **Hamming distance**:
For codewords $[f]_{FRS}$ and $[g]_{FRS}$, their Hamming distance is:
\[
d_H([f]_{FRS},[g]_{FRS})=\left|\left\{i\in[n]:\begin{bmatrix}f(a_i)\\f(\gamma a_i)\\\vdots\\f(\gamma^{s - 1}a_i)\end{bmatrix}\neq\begin{bmatrix}g(a_i)\\g(\gamma a_i)\\\vdots\\g(\gamma^{s - 1}a_i)\end{bmatrix}\right\}\right|

### Summary

By introducing the concept of piecewise divided differences, this paper successfully constructs a new coding method, enabling multivariate multiplicity codes to perform list - decoding efficiently on any finite field without depending on the characteristics of the finite field. This result not only extends the existing theory but also provides new possibilities for practical applications.