While deep learning algorithms demonstrate a great potential in scientific computing, its application to multi-scale problems remains be big challenge. This is manifested by the “frequency principle” that neural networks tend learn low frequency components first. Novel architectures such as network (MscaleDNN) were proposed alleviate this problem some extent. In paper, we construct subspace dec...

In this article, we give several generalizations of the concept of annihilating an ideal graph over a commutative ring with identity to modules. We observe that, over a commutative ring, R, AG∗(RM) is connected, and diamAG∗(RM) ≤ 3. Moreover, if AG∗(RM) contains a cycle, then grAG∗(RM) ≤ 4. Also for an R-module M with A∗(M) ̸= S(M) \ {0}, A∗(M) = ∅, if and only if M is a uniform module, and ann(...

This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of EPFL’s products or services. Internal or personal use of this material is permitted. However, permission to reprint / republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be ob...

We consider the following problem: For a system consisting of two components, the behavior of one component is known as well as the desired global behavior. What should be the behavior of the second component such that the behavior of the composition of the two conforms to the desired behavior ? This problem has been called "submodule construction" or "equation solving"; and in the context of s...