We report the first results of a light weakly interacting massive particles (WIMPs) search from the CDEX-10 experiment with a 10 kg germanium detector array immersed in liquid nitrogen at the China Jinping Underground Laboratory with a physics data size of 102.8 kg day. At an analysis threshold of 160 eVee, improved limits of 8×10^{-42} and 3×10^{-36}  cm^{2} at a 90% confidence level on spin-independent and spin-dependent WIMP-nucleon cross sections, respectively, at a WIMP mass (m_{χ}) of 5  GeV/c^{2} are achieved. The lower reach of m_{χ} is extended to 2  GeV/c^{2}.

We report results on the searches of weakly interacting massive particles (WIMPs) with sub-GeV masses (m χ) via WIMP-nucleus spin-independent scattering with Migdal effect incorporated. Analysis on time-integrated (TI) and annual modulation (AM) effects on CDEX-1B data are performed, with 737.1 kg day exposure and 160 eVee threshold for TI analysis, and 1107.5 kg day exposure and 250 eVee threshold for AM analysis. The sensitive windows in m χ are expanded by an order of magnitude to lower DM masses with Migdal effect incorporated. New limits on σ SI χN at 90% confidence level are derived as 2 × 10 −32 ∼ 7 × 10 −35 cm 2 for TI analysis at m χ ∼ 50-180 MeV=c 2 , and 3 × 10 −32 ∼ 9 × 10 −38 cm 2 for AM analysis at m χ ∼ 75 MeV=c 2-3.0 GeV=c 2 .

We report results of a search for light weakly interacting massive particle (WIMP) dark matter from the CDEX-1 experiment at the China Jinping Underground Laboratory (CJPL). Constraints on WIMP-nucleon spin-independent (SI) and spin-dependent (SD) couplings are derived with a physics threshold of 160 eVee, from an exposure of 737.1 kg-days. The SI and SD limits extend the lower reach of light WIMPs to 2 GeV and improve over our earlier bounds at WIMP mass less than 6 GeV. PACS numbers: 95.35.+d, 29.40.-n * Participating as a member of TEXONO Collaboration

We present results on light weakly interacting massive particle (WIMP) searches with annual modulation (AM) analysis on data from a 1-kg mass p-type point-contact germanium detector of the CDEX-1B experiment at the China Jinping Underground Laboratory. Datasets with a total live time of 3.2 yr within a 4.2 yr span are analyzed with analysis threshold of 250 eVee. Limits on WIMP-nucleus (χ-N ) spin-independent cross sections as function of WIMP mass (mχ) at 90% confidence level (C.L.) are derived using the dark matter halo model. Within the context of the standard halo model, the 90% C.L. allowed regions implied by the DAMA/LIBRA and CoGeNT AM-based analysis are excluded at >99.99% and 98% C.L., respectively. These results correspond to the best sensitivity at mχ<6 GeV/c 2 among WIMP AM measurements to date. PACS numbers: 95.35.+d, 98.70.Vc Compelling cosmological evidence indicates that about one-quarter of the energy density of the Universe manifests as dark matter [1], a favored candidate of which is the weakly interacting massive particle (WIMP, denoted as χ). In direct laboratory searches of WIMPs conducted with WIMP-nucleus (χ-N ) elastic scattering, positive evidence of WIMPs can only be established by assuming detailed knowledge of the background. The annual modulation (AM) analysis, on the other hand, only requires the background at the relevant energy range is stable with time. It can provide smoking-gun signatures for WIMPs independent of background modeling. Within the astrophysical dark matter halo model [2], the expected χ-N rates have distinctive AM features with maximum intensity in June and a period of 1 yr due to the Earth's motion relative to the galaxy dark matter distribution.Positive results were concluded at significance of 12.9 σ and 2.2 σ from AM-based analysis of DAMA/LIBRA [3][4][5] and CoGeNT [6-8] experiments, respectively. However, these interpretations are challenged by integrated rate experiments with liquid xenon [9-11], cryogenic bolometer [12][13][14] and ionization germanium [15][16][17][18][19] detectors, when the data were analyzed in certain scenarios where the dark matter particle properties and distributions in the Milky Way's halo are precisely defined. Comparison of AM data with differnet targets is also model dependent. The AM-allowed regions of DAMA/LIBRA and CoGeNT have been probed and excluded by AM analysis from the XMASS-1 experiment [20,21], which is limited by the diminishing sensitivities of the liquid xenon techniques at light WIMP masses (m χ ) below 6 GeV/c 2 . The and COSINE-100 [23] experiments aim to resolve this tension by a model-independent test of DAMA/LIBRA's obser-arXiv:1904.12889v2 [hep-ex]

Deep learning (DL) has emerged as an effective tool for channel estimation in wireless communication systems, especially under some imperfect environments. However, even with such unprecedented success, DL methods still serve as black boxes and the lack of explanations on their internal mechanism severely limits further improvement and extension. In this paper, we present a preliminary theoretical analysis on DL based channel estimation for multiple-antenna systems to understand and interpret its internal mechanism. Deep neural network (DNN) with rectified linear unit (ReLU) activation function is mathematically equivalent to a set of local linear functions corresponding to different input regions.Hence, the DL estimator built on it can achieve universal approximation to a large family of functions by making efficient use of piecewise linearity. We demonstrate that DL based channel estimation does not restrict to any specific signal model and will approach to the minimum mean-squared error (MMSE) estimation in various scenarios without requiring any prior knowledge of channel statistics. Therefore, DL based channel estimation outperforms or is comparable with traditional channel estimation. Simulation results confirm the accuracy of the proposed interpretation and demonstrate the effectiveness of DL based channel estimation under both linear and nonlinear signal models.

No abstract

No abstract

With the development of deep learning theory, the application of Yolov3 in fruit detection has been widely studied. Aiming at the problem that Yolov3 loses information during network transmission and the semantic feature extraction of small targets is not rich, this article proposed an improved Yolov3 cherry tomato detection algorithm. Firstly, the proposed algorithm uses dual path network as a feature extraction network to extract richer small target semantic features. Second, four feature layers with different scales are established for multiscale prediction. Finally, the improved K-means++ clustering algorithm is used to calculate the scale of anchor boxes.Experiments showed that the algorithm has a precision rate of 94.29%, a recall rate of 94.07%, and an F1 value of 94.18%. The F1 value is 1.54% higher than Faster R-CNN and 3.45% higher than Yolov3. It takes 58 ms on average to recognize an image, which provides a theoretical basis for the fruit detection.