BZPEERReading, trust and collaboration for research
Menu

Discover

GraphFollow connections around a paper, topic or saved view.

Workspaces

HomePick up where your research flow left off.InboxHandle requests, replies and notifications from one place.CollectionsCurate reading lists, evidence sets and shareable dossiers.ResearchSee your private provenance graph, trail and imports.CollaborationMove from discovery into focused discussion rooms.PublishDraft, preview and publish full-text research articles.

Network

ExpertsFind specialists worth following or asking for help.CommunitiesJoin research circles organized around shared work.PartnersSee external organizations active around the same topics.

Opportunities

ListingsBrowse roles, calls and collaborations with clearer fit signals.

Account

Log inReturn to your reading and collaboration workspace.Create accountStart saving work, following people and joining teams.
Log inCreate account

Researcher profile

Zhen Zheng

Zhen Zheng contributes to research discovery and scholarly infrastructure.

ResearcherAffiliation not importedOpen to collaborate
cond-mat.quant-gasDistributed, Parallel, and Cluster Computingcond-mat.mes-hallArtificial Intelligencecond-mat.mtrl-scicond-mat.str-el

Trust snapshot

Quick read

Trust 21 - EmergingVerification L1Unclaimed author
10works
0followers
6topics
4close collaborators

Actions

Decide how to stay connected

Follow researcher0

Identity and collaboration

How to connect with this researcher

Claiming links this public author record to a researcher profile and unlocks direct collaboration workflows.

Log in to claim

Direct collaboration

Open a focused conversation when the fit is right

Claim this author entity first to unlock direct invitations.

Research graph

See the researcher in context

Open full explorer

Inspect adjacent work, topics, institutions and collaborators without jumping out to a separate graph page.

Building this graph slice

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

10 published item(s)

preprint2022arXiv

Intertwined Weyl phases emergent from higher-order topology and unconventional Weyl fermions via crystalline symmetry

We discover three-dimensional intertwined Weyl phases, by developing a theory to create topological phases. The theory is based on intertwining existing topological gapped and gapless phases protected by the same crystalline symmetry. The intertwined Weyl phases feature both unconventional Weyl semimetallic (monopole charge>1) and higher-order topological phases, and more importantly, their exotic intertwining. While the two phases are independently stabilized by the same symmetry, their intertwining results in the specific distribution of them in the bulk. The construction mechanism allows us to combine different kinds of unconventional Weyl semimetallic and higher-order topological phases to generate distinct phases. Remarkably, on 2D surfaces, the intertwining causes the Fermi-arc topology to change in a periodic pattern against surface orientation. This feature provides a characteristic and feasible signature to probe the intertwining Weyl phases. Moreover, we provide guidelines for searching candidate materials, and elaborate on emulating the intertwined double-Weyl phase in cold-atom experiments.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Non-Hermitian spatial symmetries and their stabilized normal and exceptional topological semimetals

We study non-Hermitian spatial symmetries -- a class of symmetries that have no counterparts in Hermitian systems -- and study how normal and exceptional semimetals can be stabilized by these symmetries. Different from internal ones, spatial symmetries act nonlocally in momentum space and enforce global constraints on both band degeneracies and topological quantities at different locations. In deriving general constraints on band degeneracies and topological invariants, we demonstrate that non-Hermitian spatial symmetries are on an equal footing with, but are essentially different from Hermitian ones. First, we discover the nonlocal Hermitian conjugate pair of exceptional or normal band degeneracies that are enforced by non-Hermitian spatial symmetries. Remarkably, we find that these pairs lead to the symmetry-enforced violation of the Fermion doubling theorem in the long-time limit. Second, with the topological constraints, we unravel that certain exceptional manifold is only compatible with and stabilized by non-Hermitian spatial symmetries but is intrinsically incompatible with Hermitian spatial symmetries. We illustrate these findings using two three-dimensional models of a non-Hermitian Weyl semimetal and an exceptional unconventional Weyl semimetal. Experimental cold-atom realizations of both models are also proposed.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Quantum emulation of topological magneto-optical effects using ultracold atoms

Magneto-optical effect is a fundamental but broad concept in magnetic mediums. Here we propose a scheme for its quantum emulation using ultracold atoms. By representing the light-medium interaction in the quantum-emulation manner, the artificial magneto-optical effect emerges under an entirely different mechanism from the conventional picture. The underlying polarization state extracted in the synthetic dimension displays a different response to various experimental setups. Notably, the magneto-optical rotation is related to the bulk topology in synthetic dimensions, and thus provides an unambiguous evidence for the desired topological magneto-optical effect, which has not been developed hitherto in ultracold atoms. This scheme is simple and feasible, and can be realized by current experimental techniques. The implementation of the scheme is able to offer an intriguing platform for exploring topological magneto-optical effects and associated physics.

0 citations0 reviewsNo code linkedOpen access
preprint2022arXiv

Whale: Efficient Giant Model Training over Heterogeneous GPUs

The scaling up of deep neural networks has been demonstrated to be effective in improving model quality, but also encompasses several training challenges in terms of training efficiency, programmability, and resource adaptability. We present Whale, a general and efficient distributed training framework for giant models. To support various parallel strategies and their hybrids, Whale generalizes the programming interface by defining two new primitives in the form of model annotations, allowing for incorporating user hints. The Whale runtime utilizes those annotations and performs graph optimizations to transform a local deep learning DAG graph for distributed multi-GPU execution. Whale further introduces a novel hardware-aware parallel strategy, which improves the performance of model training on heterogeneous GPUs in a balanced manner. Deployed in a production cluster with 512 GPUs, Whale successfully trains an industry-scale multimodal model with over ten trillion model parameters, named M6, demonstrating great scalability and efficiency.

0 citations0 reviewsNo code linkedOpen access
preprint2021arXiv

Emulation of magneto-optic Faraday effect using ultracold atoms

We propose an arresting scheme for emulating the famous Faraday effect in ultracold atomic gases. Inspired by the similarities between the light field and bosonic atoms, we represent the light propagation in medium by the atomic transport in accompany of the laser-atom interaction. An artificial magneto-optic Faraday effect is readily signaled by the spin imbalance of atoms, with the setup of laser fields offering a high controllability for quantum manipulation. The present scheme is really feasible and can be realized with existing experimental techniques of ultracold atoms. It generalizes the crucial concept of the magneto-optic Faraday effect to ultracold atomic physics, and opens a new way of quantum emulating and exploring the magneto-optic Faraday effect and associated intriguing physics.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Auto-MAP: A DQN Framework for Exploring Distributed Execution Plans for DNN Workloads

The last decade has witnessed growth in the computational requirements for training deep neural networks. Current approaches (e.g., data/model parallelism, pipeline parallelism) parallelize training tasks onto multiple devices. However, these approaches always rely on specific deep learning frameworks and requires elaborate manual design, which make it difficult to maintain and share between different type of models. In this paper, we propose Auto-MAP, a framework for exploring distributed execution plans for DNN workloads, which can automatically discovering fast parallelization strategies through reinforcement learning on IR level of deep learning models. Efficient exploration remains a major challenge for reinforcement learning. We leverage DQN with task-specific pruning strategies to help efficiently explore the search space including optimized strategies. Our evaluation shows that Auto-MAP can find the optimal solution in two hours, while achieving better throughput on several NLP and convolution models.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Cavity-induced Fulde-Ferrell-Larkin-Ovchinnikov superfluids of ultracold Fermi gases

Motivated by recent experimental advances in ultracold atomic gases placed in cavities, we study the influence of the atom-cavity coupling on the Fermi gases trapped in optical lattices. By adiabatic elimination of the cavity photon field, the atom-cavity coupling gives rise to effective long-range interactions. It results in a variety of two-body scattering processes, during which the atomic pairs can acquire an additional center-of-mass momentum. This reveals the possibility of Fulde-Ferrell-Larkin-Ovchinnikov (FFLO) superfluids in which the atomic pairing momentum is nonzero. By inspecting the phase diagram at the mean-field level, we confirm that the FFLO superfluid phase coexists with the zero-momentum pairing, and is the ground state that hosts the lowest energy. Furthermore, the order parameter characterizing the nonzero-momentum pairing does not vanish as long as the cavity-induced interaction is present.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

DAPPLE: A Pipelined Data Parallel Approach for Training Large Models

It is a challenging task to train large DNN models on sophisticated GPU platforms with diversified interconnect capabilities. Recently, pipelined training has been proposed as an effective approach for improving device utilization. However, there are still several tricky issues to address: improving computing efficiency while ensuring convergence, and reducing memory usage without incurring additional computing costs. We propose DAPPLE, a synchronous training framework which combines data parallelism and pipeline parallelism for large DNN models. It features a novel parallelization strategy planner to solve the partition and placement problems, and explores the optimal hybrid strategy of data and pipeline parallelism. We also propose a new runtime scheduling algorithm to reduce device memory usage, which is orthogonal to re-computation approach and does not come at the expense of training throughput. Experiments show that DAPPLE planner consistently outperforms strategies generated by PipeDream's planner by up to 3.23x under synchronous training scenarios, and DAPPLE runtime outperforms GPipe by 1.6x speedup of training throughput and reduces the memory consumption of 12% at the same time.

0 citations0 reviewsNo code linkedOpen access
preprint2020arXiv

Effective Hamiltonian with tunable mixed pairing in driven optical lattices

Mixed pairing in ultracold Fermi gases can give rise to interesting many-body phases, such as topological nontrivial superfluids that support Majorana zero modes (MZMs) with various spatial configurations. Unfortunately, in ordinary lattice systems, the topological phase and the associated MZMs are suppressed by the dominant $s$-wave pairing. Here we present a proposal for engineering effective Hamiltonians with tunable mixed on- and off-site pairing based on driven optical lattices. The on- and off-site pairing can be changed independently by means of a periodical driving field rather than magnetic Feshbach resonances. It paves the way for suppressing the dominant on-site interaction that frustrates the emergence of topological superfluids and for synthesizing MZMs localized in edges or corners.

0 citations0 reviewsNo code linkedOpen access
preprint2017arXiv

Topological Supersolidity of Dipolar Fermi Gases in a Spin-Dependent Optical Lattice

We investigate topological supersolidity of dipolar Fermi gases in a spin-dependent 2D optical lattice. Numerical results show that the topological supersolid states can be synthesized via the combination of topological superfluid states with the stripe order, where the topological superfluid states generated with dipolar interaction possess the $Δ_{x}+iΔ_{y}$ order, and it is of D class topological classification. By adjusting the ratio between hopping amplitude $t_{x}/t_{y}$ and interaction strength $U$ with dipole orientation $ϕ\approx \fracπ{4}$, the system will undergo phase transitions among the $p_{x}+ip_{y}$-wave topological superfluid state, the p-wave superfluid state, and the topological supersolid state. The topological supersolid state is proved to be stable by the positive sign of the inverse compressibility. We design an experimental protocol to realize the staggered next-next-nearest-neighbour hopping via the laser assisted tunneling technique, which is the key to synthesize topological supersolid states.

0 citations0 reviewsNo code linkedOpen access