Xiang Zhang

Bio: Xiang Zhang is an academic researcher from Baylor College of Medicine. The author has contributed to research in topics: Medicine & Computer science. The author has an hindex of 154, co-authored 1733 publications receiving 117576 citations. Previous affiliations of Xiang Zhang include University of California, Berkeley & University of Texas MD Anderson Cancer Center.

Space-Time Crystals of Trapped Ions

Abstract: Spontaneous symmetry breaking can lead to the formation of time crystals, as well as spatial crystals. Here we propose a space-time crystal of trapped ions and a method to realize it experimentally by confining ions in a ring-shaped trapping potential with a static magnetic field. The ions spontaneously form a spatial ring crystal due to Coulomb repulsion. This ion crystal can rotate persistently at the lowest quantum energy state in magnetic fields with fractional fluxes. The persistent rotation of trapped ions produces the temporal order, leading to the formation of a space-time crystal. We show that these space-time crystals are robust for direct experimental observation. We also study the effects of finite temperatures on the persistent rotation. The proposed space-time crystals of trapped ions provide a new dimension for exploring many-body physics and emerging properties of matter.

A carpet cloak for visible light.

Abstract: We report an invisibility carpet cloak device, which is capable of making an object undetectable by visible light. The cloak is designed using quasi conformal mapping and is fabricated in a silicon nitride waveguide on a specially developed nanoporous silicon oxide substrate with a very low refractive index (n<1.25). The spatial index variation is realized by etching holes of various sizes in the nitride layer at deep subwavelength scale creating a local effective medium index. The fabricated device demonstrates wideband invisibility throughout the visible spectrum with low loss. This silicon nitride on low index substrate can also be a general scheme for implementation of transformation optical devices at visible frequencies.

Rapidity and transverse-momentum dependence of the inclusive J/psi nuclear modification factor in p-Pb collisions at root s(NN)=5.02 TeV

Abstract: We have studied the transverse-momentum (pT) dependence of the inclusive J/ψ production in p-Pb collisions at √sNN = 5.02 TeV, in three center-of-mass rapidity (ycms) regions, down to zero pT. Results in the forward and backward rapidity ranges (2.03 < ycms < 3.53 and −4.46 < ycms < −2.96) are obtained by studying the J/ψ decay to µ +µ −, while the mid-rapidity region (−1.37 < ycms < 0.43) is investigated by measuring the e+e − decay channel. The pT dependence of the J/ψ production cross section and nuclear modification factor are presented for each of the rapidity intervals, as well as the J/ψ mean pT values. Forward and mid-rapidity results show a suppression of the J/ψ yield, with respect to pp collisions, which decreases with increasing pT. At backward rapidity no significant J/ψ suppression is observed. Theoretical models including a combination of cold nuclear matter effects such as shadowing and partonic energy loss, are in fair agreement with the data, except at forward rapidity and low transverse momentum. The implications of the p-Pb results for the evaluation of cold nuclear matter effects on J/ψ production in Pb-Pb collisions are also discussed.

Strange particle production in proton-proton collisions at root s=0.9 TeV with ALICE at the LHC

Abstract: The production of mesons containing strange quarks ($\Kzs$, $\phi$) and both singly and doubly strange baryons ($\rmLambda$, $\rmAlambda$, and $\Xis$) are measured at central rapidity in pp collisions at $\sqrt{s}$ = 0.9 $\tev$ with the ALICE experiment at the LHC. The results are obtained from the analysis of about 250 k minimum bias events recorded in 2009. Measurements of yields (dN/dy) and transverse momentum spectra at central rapidities for inelastic pp collisions are presented. For mesons, we report yields ($ $) of $0.184 \pm 0.002 \stat \pm 0.006 \syst$ for $\Kzs$ and $0.021 \pm 0.004 \stat \pm 0.003 \syst$ for $\phi$. For baryons, we find $ = 0.048 \pm 0.001 \stat \pm 0.004 \syst$ for $\rmLambda$, $0.047 \pm 0.002 \stat \pm 0.005 \syst$ for $\rmAlambda$ and $0.0101 \pm 0.0020 \stat \pm 0.0009 \syst$ for $\Xis$. The results are also compared with predictions for identified particle spectra from QCD-inspired models and provide a baseline for comparisons with both future pp measurements at higher energies and heavy-ion collisions.

Immuno-subtyping of breast cancer reveals distinct myeloid cell profiles and immunotherapy resistance mechanisms.

Abstract: Cancer-induced immune responses affect tumour progression and therapeutic response. In multiple murine models and clinical datasets, we identified large variations of neutrophils and macrophages that define 'immune subtypes' of triple-negative breast cancer (TNBC), including neutrophil-enriched (NES) and macrophage-enriched subtypes (MES). Different tumour-intrinsic pathways and mutual regulation between macrophages (or monocytes) and neutrophils contribute to the development of a dichotomous myeloid compartment. MES contains predominantly macrophages that are CCR2-dependent and exhibit variable responses to immune checkpoint blockade (ICB). NES exhibits systemic and local accumulation of immunosuppressive neutrophils (or granulocytic myeloid-derived suppressor cells), is resistant to ICB, and contains a minority of macrophages that seem to be unaffected by CCR2 knockout. A MES-to-NES conversion mediated acquired ICB resistance of initially sensitive MES models. Our results demonstrate diverse myeloid cell frequencies, functionality and potential roles in immunotherapies, and highlight the need to better understand the inter-patient heterogeneity of the myeloid compartment.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Going deeper with convolutions

Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.