Searcharxiv⌕ Search

arXiv subjects

Yi Sun

Publications and source records attributed to Yi Sun.

192 records · Page 11Linked to original sources
arXiv2011

Observation of Landau level-like quantizations at 77 K along a strained-induced graphene ridge

Recent studies show that the electronic structures of graphene can be modified by strain and it was predicted that strain in graphene can induce peaks in the local density of states (LDOS) mimicking Landau levels (LLs) generated in the presence of a large magnetic field. Here we report scanning tunnelling spectroscopy (STS) observation of nine strain-induced peaks in LDOS at 77 K along a graphene ridge created when the graphene layer was cleaved from a sample of highly oriented pyrolytic graphite (HOPG). The energies of these peaks follow the progression of LLs of massless 'Dirac fermions' (DFs) in a magnetic field of 230 T. The results presented here suggest a possible route to realize zero-field quantum Hall-like effects at 77 K.

cond-mat.str-el↗
arXiv2011

Natural Evolution Strategies

This paper presents Natural Evolution Strategies (NES), a recent family of algorithms that constitute a more principled approach to black-box optimization than established evolutionary algorithms. NES maintains a parameterized distribution on the set of solution candidates, and the natural gradient is used to update the distribution's parameters in the direction of higher expected fitness. We introduce a collection of techniques that address issues of convergence, robustness, sample complexity, computational complexity and sensitivity to hyperparameters. This paper explores a number of implementations of the NES family, ranging from general-purpose multi-variate normal distributions to heavy-tailed and separable distributions tailored towards global optimization and search in high dimensional spaces, respectively. Experimental results show best published performance on various standard benchmarks, as well as competitive performance on others.

stat.ML↗
arXiv2011

Effect of Exchange-type Zero-bias Anomaly on Single Electron Tunnelling of Au Nanoparticles

Using cryogenic scanning tunnelling microscopy and scanning tunnelling spectroscopy we measured single electron tunnelling of isolated Au nanoparticles with 1.4 nm in radius. We observe that a gap ΔV ~ 2e/C (C is the capacitance of the Au particle) around zero bias in the tunnelling conductance spectrum, followed by a series of discrete single electron tunnelling peaks with voltage widths of EC ~ e/C at both negative and positive bias. Experimental data are well explained by taking into account the effect of exchange interaction of electrons on the single electron tunnelling of Au nanoparticles. A tunnelling peak near zero-bias was suppressed by the exchange-type zero-bias anomaly, which results in the gap ΔV ~ 2EC.

cond-mat.mes-hall↗
arXiv2011

A Linear Time Natural Evolution Strategy for Non-Separable Functions

We present a novel Natural Evolution Strategy (NES) variant, the Rank-One NES (R1-NES), which uses a low rank approximation of the search distribution covariance matrix. The algorithm allows computation of the natural gradient with cost linear in the dimensionality of the parameter space, and excels in solving high-dimensional non-separable problems, including the best result to date on the Rosenbrock function (512 dimensions).

cs.AI↗
arXiv2011

Axiomatic Attribution for Multilinear Functions

We study the attribution problem, that is, the problem of attributing a change in the value of a characteristic function to its independent variables. We make three contributions. First, we propose a formalization of the problem based on a standard cost sharing model. Second, we show that there is a unique attribution method that satisfies Dummy, Additivity, Conditional Nonnegativity, Affine Scale Invariance, and Anonymity for all characteristic functions that are the sum of a multilinear function and an additive function. We term this the Aumann-Shapley-Shubik method. Conversely, we show that such a uniqueness result does not hold for characteristic functions outside this class. Third, we study multilinear characteristic functions in detail; we describe a computationally efficient implementation of the Aumann-Shapley-Shubik method and discuss practical applications to pay-per-click advertising and portfolio analysis.

cs.GT↗
arXiv2008

Quasi-Large Sparse-Sequence CDMA: Approach to Single-User Bound by Linearly-Complex LAS Detectors

We have proposed a quasi-large random-sequence (QLRS) CDMA where K users access a point through a common channel with spectral spreading factor N. Each bit is extended by a temporal spreading factor B and hopped on a BN-chip random sequence that is spread in time and frequency. Each user multiplexes and transmits B extended bits and the total channel load is alpha = K/N bits/s/Hz. The linearly-complex LAS detectors detect the transmitted bits. We have obtained that as B tends to infinity, if alpha < 1/2 - 1/(4ln2), each transmitted bit achieves the single-bit bound in BER in high SNR regime as if there was no interference bit. In simulation, when bit number BK >= 500, each bit can approach the single-bit bound for alpha as high as 1 bit/s/Hz. In this paper, we further propose the quasi-large sparse-sequence (QLSS) CDMA by replacing the dense sequence in QLRS-CDMA with sparse sequence. Simulation results show that when the nonzero chips are as few as 16, the BER is already near that of QLRS-CDMA while the complexity is significantly reduced due to sequence sparsity.

cs.IT↗
arXiv2008

A Family of Likelihood Ascent Search Multiuser Detectors: an Upper Bound of Bit Error Rate and a Lower Bound of Asymptotic Multiuser Efficiency

In this paper, the bit error performance of a family of likelihood ascent search (LAS) multiuser detectors is analyzed. An upper bound on the BER of any LAS detector is obtained by bounding the fixed point region with the worst initial detector. The concept of indecomposable errors developed by Verdu is applied to tighten the upper bound. In a special instance, the upper bound is reduced to that for all the local maximum likelihood detectors. The upper bound is comparable with that of the optimum detector obtained by Verdu. A lower bound on the asymptotic multiuser efficiency (AME) is then obtained. It is shown that there are nontrivial CDMA channels such that a LAS detector can achieve unit AME regardless of user number. The AME lower bound provides a means for further seeking a good set of spreading sequences and power distribution for spectral and power efficient CDMA.

cs.IT↗
arXiv2007

A Family of Likelihood Ascent Search Multiuser Detectors: Approach to Single-User Performance via Quasi-Large Random Sequence CDMA

Since Tse and Verdu proved that the global maximum likelihood (GML) detector achieves unit asymptotic multiuser efficiency (AME) in the limit of large random spreading (LRS) CDMA, no suboptimal detector has been found to achieve unit AME. In this letter, we obtain that the WSLAS detector with a linear per-bit complexity achieves unit AME in the LRS-CDMA with a channel load < 1/2 - 1/(4ln2) bits/s/Hz. For a practical system with any user number, a quasi LRS-CDMA is then proposed to approach the single-user performance in the high SNR regime.

cs.IT↗
arXiv1994

Calculation of Gromov-Witten invariants for $ CP^{3},CP^{4}$,and $Gr(2,4)$

Using the associativity relations of the topological Sigma Models with target spaces, $CP^3, CP^4$ and $Gr(2,4)$ , we derive recursion relations of their correlation and evaluate them up to certain order in the expansion over the instantons. The expansion coeffieients are regarded as the number of rational curves in $CP^3, CP^4$ and $Gr(2,4)$ which intersect various types of submanifolds corresponding to the choice of BRST invariant operators in the correlation functions.

hep-th↗