Searcharxiv⌕ Search

arXiv subjects

Songsong Huang

Publications and source records attributed to Songsong Huang.

4 recordsLinked to original sources

GV-subschemes and hyperelliptic Jacobians

We provide a solution to the hyperelliptic Schottky problem: a principally polarized complex abelian variety whose theta divisor has singular locus of codimension three is a hyperelliptic Jacobian. As a key step towards this criterion for hyperelliptic Jacobians, we prove a conjecture of Pareschi and Popa which predicts that if the ideal sheaf $I_X$ of a subscheme $X$ of an indecomposable PPAV $(A, Θ)$ satisfies that $I_X(2Θ)$ is IT$_0$, then $X$ is a GV-subscheme. As a corollary of the main result, the base scheme of an ample line bundle on a complex abelian variety is generically reduced along every irreducible component of codimension $2$, confirming a conjecture of Debarre. This corollary completes the algebro-geometry proof by Arbarello-Codogni-Pareschi in \cite{ACP21, ACP25} of Novikov's conjecture, which characterizes Jacobians via the KP equation.

math.AG↗

Almgren minimality of products of simplex-skeleton cones

In this paper, we prove that for every $m,n\geq 2$, the orthogonal product $S_m\times S_n\subset\R^m\times\R^n$ is an Almgren minimal cone of dimension $m+n-2$, where $S_k\subset\R^k$ is the cone over the $(k-2)$-skeleton of a regular $k$-simplex. Lawlor and Morgan \cite{LM94} proved the minimality of each $S_k$ by paired calibrations, while Guy David \cite{DJT, GD13} later raised the general question whether products of Almgren minimal sets remain minimal. Our theorem gives a affirmative answer to this product question within the entire Lawlor-Morgan regular-simplex family. It extends the isolated example \(\Y\times \Y\), the product of two 1-dimensional $\Y$ sets, proved in 2014 by the second author, simultaneously to all pairwise products \(S_m\times S_n\), and therefore produces a large new family of possible codimension-two singularities for Plateau's problem. The key idea of the proof is to find a new way to combine homology with paired calibrations in higher codimension. Instead of using only the currents associated with the facets, we consider all their nonzero $\FF$ combinations. The resulting complete family has enough algebraic symmetry to control uniformly the overlaps created by noninjective deformations.

math.CA↗

Degree of irrationality of abelian surfaces

Let $A$ be a complex abelian surface. We prove that $A$ admits a dominant rational map of degree three to $\PP^2$ if and only if it admits a polarization of type $(1,1)$, $(1,2)$, $(1,3)$, or $(1,6)$. Consequently, the degree of irrationality of $A$ is $3$ precisely in these four cases and is $4$ otherwise.

math.AG↗

Learn To Learn More Precisely

Meta-learning has been extensively applied in the domains of few-shot learning and fast adaptation, achieving remarkable performance. While Meta-learning methods like Model-Agnostic Meta-Learning (MAML) and its variants provide a good set of initial parameters for the model, the model still tends to learn shortcut features, which leads to poor generalization. In this paper, we propose the formal conception of "learn to learn more precisely", which aims to make the model learn precise target knowledge from data and reduce the effect of noisy knowledge, such as background and noise. To achieve this target, we proposed a simple and effective meta-learning framework named Meta Self-Distillation(MSD) to maximize the consistency of learned knowledge, enhancing the models' ability to learn precise target knowledge. In the inner loop, MSD uses different augmented views of the same support data to update the model respectively. Then in the outer loop, MSD utilizes the same query data to optimize the consistency of learned knowledge, enhancing the model's ability to learn more precisely. Our experiment demonstrates that MSD exhibits remarkable performance in few-shot classification tasks in both standard and augmented scenarios, effectively boosting the accuracy and consistency of knowledge learned by the model.

cs.LG↗