Searcharxiv⌕ Search

arXiv subjects

Li Jiang

Publications and source records attributed to Li Jiang.

137 records · Page 8Linked to original sources
arXiv2018

Approximate Random Dropout

The training phases of Deep neural network~(DNN) consumes enormous processing time and energy. Compression techniques utilizing the sparsity of DNNs can effectively accelerate the inference phase of DNNs. However, it can be hardly used in the training phase because the training phase involves dense matrix-multiplication using General Purpose Computation on Graphics Processors (GPGPU), which endorse regular and structural data layout. In this paper, we propose the Approximate Random Dropout that replaces the conventional random dropout of neurons and synapses with a regular and predefined patterns to eliminate the unnecessary computation and data access. To compensate the potential performance loss we develop a SGD-based Search Algorithm to produce the distribution of dropout patterns. We prove our approach is statistically equivalent to the previous dropout method. Experiments results on MLP and LSTM using well-known benchmarks show that the proposed Approximate Random Dropout can reduce the training time by $20\%$-$77\%$ ($19\%$-$60\%$) when dropout rate is $0.3$-$0.7$ on MLP (LSTM) with marginal accuracy drop.

cs.LG↗
arXiv2018

Invocation-driven Neural Approximate Computing with a Multiclass-Classifier and Multiple Approximators

Neural approximate computing gains enormous energy-efficiency at the cost of tolerable quality-loss. A neural approximator can map the input data to output while a classifier determines whether the input data are safe to approximate with quality guarantee. However, existing works cannot maximize the invocation of the approximator, resulting in limited speedup and energy saving. By exploring the mapping space of those target functions, in this paper, we observe a nonuniform distribution of the approximation error incurred by the same approximator. We thus propose a novel approximate computing architecture with a Multiclass-Classifier and Multiple Approximators (MCMA). These approximators have identical network topologies and thus can share the same hardware resource in a neural processing unit(NPU) clip. In the runtime, MCMA can swap in the invoked approximator by merely shipping the synapse weights from the on-chip memory to the buffers near MAC within a cycle. We also propose efficient co-training methods for such MCMA architecture. Experimental results show a more substantial invocation of MCMA as well as the gain of energy-efficiency.

cs.LG↗
arXiv2018

Angle-dependent magic wavelengths for the $4s_{1/2}\to3d_{5/2,3/2}$ transitions of Ca$^{+}$ ions

The dynamic polarizabilities of the atomic states with angular momentum $j> \frac12$ are sensitive to the angle between the quantization axis $\hat{e}_z$ and the polarization vector $\hat{\mathbfε}$ owing to the contribution of anisotropic tensor polarizabilities. The magic wavelength, at which the differential Stark shift of an atomic transition nullifies, depends on this angle. We identified the magic wavelengths for the $4s_{\frac12}\to3d_{\frac32,\frac52}$ transitions of Ca$^{+}$ ions at different angles between $\hat{e}_z$ and $\hat{\mathbfε}$ in the case of linearly polarized light. We found that the magic wavelengths near 395.79 nm, which lie between the $4s_{\frac12}\to4p_{\frac12}$ and $4s_{\frac12}\to 4p_{\frac32}$ transition wavelengths, remain unsensitive to the angle, while the magic wavelengths, which are longer than the $3d_{\frac52}\to 4p_{\frac32}$ resonant transition wavelength (854.21 nm), are very sensitive to the angle.

physics.atom-ph↗
arXiv2017

Informative and misinformative interactions in a school of fish

It is generally accepted that, when moving in groups, animals process information to coordinate their motion. Recent studies have begun to apply rigorous methods based on Information Theory to quantify such distributed computation. Following this perspective, we use transfer entropy to quantify dynamic information flows locally in space and time across a school of fish during directional changes around a circular tank, i.e. U-turns. This analysis reveals peaks in information flows during collective U-turns and identifies two different flows: an informative flow (positive transfer entropy) based on fish that have already turned about fish that are turning, and a misinformative flow (negative transfer entropy) based on fish that have not turned yet about fish that are turning. We also reveal that the information flows are related to relative position and alignment between fish, and identify spatial patterns of information and misinformation cascades. This study offers several methodological contributions and we expect further application of these methodologies to reveal intricacies of self-organisation in other animal groups and active matter in general.

q-bio.QM↗
arXiv2017

Magic wavelengths of Ca$^{+}$ ion for linearly and circularly polarized light

The dynamic dipole polarizabilities of the low-lying states of Ca$^{+}$ for linearly and circularly polarized light are calculated by using relativistic configuration interaction plus core polarization (RCICP) approach. The magic wavelengths, at which the two levels of the transitions have the same ac Stark shifts, for $4s$-$4p_{j,m}$ and $4s$-$3d_{j,m}$ magnetic sublevels transitions are determined. The present magic wavelengths for linearly polarized light agree with the available results excellently. The polarizability for the circularly polarized light has the scalar, vector and tensor components. The dynamic polarizability is different for each of magnetic sublevels of the atomic state. Additional magic wavelengths have been found for the circularly polarized light. We recommend that the measurement of the magic wavelength near 850 nm for $4s-4p_{\frac32,m=\pm\frac32,\pm\frac12}$ could be able to determine the oscillator strength ratio of $f_{4p_{\frac32} \to 3d_{\frac32}}$ and $f_{4p_{\frac32} \to 3d_{\frac52}}$.

physics.atom-ph↗
arXiv2017

Non-Markovian dynamics for a two-atom-coupled system interacting with local reservoir at finite temperature

By using the effective non-Markovian measure [H.P. Breuer, E.M. Laine, J. Piilo, Phys. Rev. Lett. 103, 210401 (2009)], we investigate non-Markovian dynamics of a pair of two-level atoms (TLAs) system, each of which interacting with a local reservoir. We show that subsystem dynamics can be controlled by manipulating the coupling between TLAs, temperature and relaxation rate of the atoms. Moreover, the correlation between non-Markovianity of subsystem and entanglement between the subsystem and the structured bath is investigated, the results show that the emergence of non-Markovianity has a negative effect on the entanglement.

quant-ph↗
arXiv2017

Electric Switching of Fluorescence Decay in Gold-Silica-Dye Nematic Nanocolloids Mediated by Surface Plasmons

Tunable composite materials with interesting physical behavior can be designed through integrating unique optical properties of solid nanostructures with the facile responses of soft matter to weak external stimuli, but this approach remains challenged by their poorly controlled co-assembly at the mesoscale. Using scalable wet chemical synthesis procedures, we fabricated anisotropic gold-silica-dye colloidal nanostructures and then organized them into the device-scale (demonstrated for square inch cells) electrically tunable composites by simultaneously invoking molecular and colloidal self-assembly. We show that the ensuing ordered colloidal dispersions of shape-anisotropic nanostructures exhibit tunable fluorescence decay rates and intensity. We characterize how these properties depend on low-voltage fields and polarization of both the excitation and emission light, demonstrating a great potential for the practical realization of an interesting breed of nanostructured composite materials.

cond-mat.soft↗
arXiv2015

Method for measuring prompt gamma-rays generated by D-T neutrons bombarding a depleted uranium spherical shell

The prompt gamma-ray spectrum from depleted uranium (DU) spherical shells induced by 14 MeV D-T neutrons is measured. Monte Carlo (MC) simulation gives the largest prompt gamma flux with the optimal thickness of the DU spherical shells 3-5 cm and the optimal frequency of neutron pulse 1 MHz. The method of time of flight and pulse shape coincidence with energy (DC-TOF) is proposed, and the subtraction of the background gamma-rays discussed in detail. The electron recoil spectrum and time spectrum of the prompt gamma-rays are obtained based on a 2"*2" BC501A liquid scintillator detector. The energy spectrum and time spectrum of prompt gamma-rays are obtained based on an iterative unfolding method that can remove the influence of γ-rays response matrix and pulsed neutron shape. The measured time spectrum and the calculated results are roughly consistent with each other. Experimental prompt gamma-ray spectrum in the 0.4-3 MeV energy region agree well with MC simulation based on the ENDF/BVI.5 library, and the discrepancies for the integral quantities of gamma-rays of energy 0.4-1 MeV and 1-3 MeV are 9.2% and 1.1%, respectively.

physics.ins-det↗
arXiv2002

Interacting Dipoles from Matrix Formulation of Noncommutative Gauge Theories

We study the IR behavior of noncommutative gauge theory in the matrix formulation. We find that in this approach, the nature of the UV/IR mixing is easily understood, which allows us to perform a reliable calculation of the quantum effective action for the long wavelength modes of the noncommutative gauge field. At one loop, we find that our description is weakly coupled only in the supersymmetric theory. At two loops, we find non-trivial interaction terms suggestive of dipole degrees of freedom. These dipoles exhibit a channel duality reminiscent of string theory.

hep-th↗
arXiv2000

Dirac Monopole in Non-Commutative Space

We consider static U(1) monopole in non-commutative space. Up to the second order in the non-commutativity scale $θ$, we find no non-trivial corrections to the Dirac solution, the monopole mass remains infinite. We argue the same holds for any arbitrary higher order. Some speculation about the nature of non-commutative spacetime and its relation to the cosmological constant is made.

hep-th↗