Flow coefficients v_{n} for n=2, 3, 4, characterizing the anisotropic collective flow in Au+Au collisions at sqrt[s_{NN}]=200  GeV, are measured relative to event planes Ψ_{n}, determined at large rapidity. We report v_{n} as a function of transverse momentum and collision centrality, and study the correlations among the event planes of different order n. The v_{n} are well described by hydrodynamic models which employ a Glauber Monte Carlo initial state geometry with fluctuations, providing additional constraining power on the interplay between initial conditions and the effects of viscosity as the system evolves. This new constraint can serve to improve the precision of the extracted shear viscosity to entropy density ratio η/s.

Back-to-back hadron pair yields in d+Au and p+p collisions at sqrt[s_{NN}]=200  GeV were measured with the PHENIX detector at the Relativistic Heavy Ion Collider. Rapidity separated hadron pairs were detected with the trigger hadron at pseudorapidity |η|<0.35 and the associated hadron at forward rapidity (deuteron direction, 3.0<η<3.8). Pairs were also detected with both hadrons measured at forward rapidity; in this case, the yield of back-to-back hadron pairs in d+Au collisions with small impact parameters is observed to be suppressed by a factor of 10 relative to p+p collisions. The kinematics of these pairs is expected to probe partons in the Au nucleus with a low fraction x of the nucleon momenta, where the gluon densities rise sharply. The observed suppression as a function of nuclear thickness, p_{T}, and η points to cold nuclear matter effects arising at high parton densities.

We present measurements of J/ψ yields in d+Au collisions at sqrt[s_{NN}]=200  GeV recorded by the PHENIX experiment and compare them with yields in p+p collisions at the same energy per nucleon-nucleon collision. The measurements cover a large kinematic range in J/ψ rapidity (-2.2<y<2.4) with high statistical precision and are compared with two theoretical models: one with nuclear shadowing combined with final state breakup and one with coherent gluon saturation effects. In order to remove model dependent systematic uncertainties we also compare the data to a simple geometric model. The forward rapidity data are inconsistent with nuclear modifications that are linear or exponential in the density weighted longitudinal thickness, such as those from the final state breakup of the bound state.

Automated scene interpretation has benefited from advances in machine learning, and restricted tasks, such as face detection, have been solved with sufficient accuracy for restricted settings. However, the performance of machines in providing rich semantic descriptions of natural scenes from digital images remains highly limited and hugely inferior to that of humans. Here we quantify this "semantic gap" in a particular setting: We compare the efficiency of human and machine learning in assigning an image to one of two categories determined by the spatial arrangement of constituent parts. The images are not real, but the category-defining rules reflect the compositional structure of real images and the type of "reasoning" that appears to be necessary for semantic parsing. Experiments demonstrate that human subjects grasp the separating principles from a handful of examples, whereas the error rates of computer programs fluctuate wildly and remain far behind that of humans even after exposure to thousands of examples. These observations lend support to current trends in computer vision such as integrating machine learning with parts-based modeling.

We propose a new learning strategy for object detection. The proposed scheme forgoes the need to train a collection of detectors dedicated to homogeneous families of poses, and instead learns a single classifier that has the inherent ability to deform based on the signal of interest.We train a detector with a standard AdaBoost procedure by using combinations of pose-indexed features and pose estimators. This allows the learning process to select and combine various estimates of the pose with features able to compensate for variations in pose without the need to label data for training or explore the pose space in testing. We validate our framework on three types of data: hand video sequences, aerial images of cars as well as face images. We compare our method to a standard boosting framework, with access to the same ground truth, and show a reduction in the false alarm rate of up to an order of magnitude. Where possible, we compare our method to the state-of-the art, which requires pose annotations of the training data, and demonstrate comparable performance.

Large parity-violating longitudinal single-spin asymmetries A(L)(e+)=-0.86(-0.14)(+0.30) and A(L)(e-)= 0.88(-0.71)(+0.12) are observed for inclusive high transverse momentum electrons and positrons in polarized p+p collisions at a center-of-mass energy of sqrt[s]= 500 GeV with the PHENIX detector at RHIC. These e± come mainly from the decay of W± and Z0 bosons, and their asymmetries directly demonstrate parity violation in the couplings of the W± to the light quarks. The observed electron and positron yields were used to estimate W± boson production cross sections for the e± channels of σ(pp → W+ X) × BR(W+ → e+ ν(e))= 144.1 ± 21.2(stat)(-10.3)(+3.4)(syst) ± 21.6(norm)  pb, and σ(pp → W- X) × BR(W- → e- ν[over ¯](e))= 31.7 ± 12.1(stat)(-8.2)(+10.1)(syst) ± 4.8(norm)  pb.

Multi-object tracking can be achieved by detecting objects in individual frames and then linking detections across frames. Such an approach can be made very robust to the occasional detection failure: If an object is not detected in a frame but is in previous and following ones, a correct trajectory will nevertheless be produced. By contrast, a false-positive detection in a few frames will be ignored. However, when dealing with a multiple target problem, the linking step results in a difficult optimization problem in the space of all possible families of trajectories. This is usually dealt with by sampling or greedy search based on variants of Dynamic Programming, which can easily miss the global optimum. In this paper, we show that reformulating that step as a constrained flow optimization results in a convex problem. We take advantage of its particular structure to solve it using the k-shortest paths algorithm, which is very fast. This new approach is far simpler formally and algorithmically than existing techniques and lets us demonstrate excellent performance in two very different contexts.

The production of e;{+}e;{-} pairs for m_{e;{+}e;{-}}<0.3 GeV/c;{2} and 1<p_{T}<5 GeV/c is measured in p+p and Au+Au collisions at sqrt[s_{NN}]=200 GeV. An enhanced yield above hadronic sources is observed. Treating the excess as photon internal conversions, the invariant yield of direct photons is deduced. In central Au+Au collisions, the excess of the direct photon yield over p+p is exponential in transverse momentum, with an inverse slope T=221+/-19;{stat}+/-19;{syst} MeV. Hydrodynamical models with initial temperatures ranging from T_{init} approximately 300-600 MeV at times of approximately 0.6-0.15 fm/c after the collision are in qualitative agreement with the data. Lattice QCD predicts a phase transition to quark gluon plasma at approximately 170 MeV.

Most state-of-the-art algorithms for filament detection in 3-D image-stacks rely on computing the Hessian matrix around individual pixels and labeling these pixels according to its eigenvalues. This approach, while very effective for clean data in which linear structures are nearly cylindrical, loses its effectiveness in the presence of noisy data and irregular structures. In this paper, we show that using steerable filters to create rotationally invariant features that include higher-order derivatives and training a classifier based on these features lets us handle such irregular structures. This can be done reliably and at acceptable computational cost and yields better results than state-of-the-art methods.

Bose-Einstein correlations of charged kaons are used to probe Au+Au collisions at sqrt[s_{NN}]=200 GeV and are compared to charged pion probes, which have a larger hadronic scattering cross section. Three-dimensional Gaussian source radii are extracted, along with a one-dimensional kaon emission source function. The centrality dependences of the three Gaussian radii are well described by a single linear function of N_{part};{1/3} with a zero intercept. Imaging analysis shows a deviation from a Gaussian tail at r greater, similar10 fm, although the bulk emission at lower radius is well described by a Gaussian. The presence of a non-Gaussian tail in the kaon source reaffirms that the particle emission region in a heavy-ion collision is extended, and that similar measurements with pions are not solely due to the decay of long-lived resonances.