Abstract. Bayesian belief nets (BNs) are often used for classification tasks -typically to return the most likely class label for each specified instance. Many BN-learners, however, attempt to find the BN that maximizes a different objective function -viz., likelihood, rather than classification accuracy -typically by first learning an appropriate graphical structure, then finding the parameters for that structure that maximize the likelihood of the data. As these parameters may not maximize the classification accuracy, "discriminative parameter learners" follow the alternative approach of seeking the parameters that maximize conditional likelihood (CL), over the distribution of instances the BN will have to classify. This paper first formally specifies this task, shows how it extends standard logistic regression, and analyzes its inherent sample and computational complexity. We then present a general algorithm for this task, ELR, that applies to arbitrary BN structures and that works effectively even when given incomplete training data. Unfortunately, ELR is not guaranteed to find the parameters that optimize conditional likelihood; moreover, even the optimal-CL parameters need not have minimal classification error. This paper therefore presents empirical evidence that ELR produces effective classifiers, often superior to the ones produced by the standard "generative" algorithms, especially in common situations where the given BN-structure is incorrect.

The STAR CollaborationNuclear collisions recreate conditions in the universe microseconds after the Big Bang. Only a very small fraction of the emitted fragments are light nuclei, but these states are of fundamental interest. We report the observation of antihypertritons -composed of an antiproton, antineutron, and antilambda hyperon -produced by colliding gold nuclei at high energy. Our analysis yields 70 ± 17 antihypertritons ( Nuclei are abundant in the universe, but antinuclei that are heavier than the antiproton have been observed only as products of interactions at particle accelerators (1, 2). Collisions of heavy nuclei at the Relativistic Heavy-Ion Collider (RHIC) at Brookhaven National Laboratory (BNL) briefly produce hot and dense matter that has been interpreted as a quark gluon plasma (QGP) (3, 4) with an energy density similar to that of the universe a few microseconds after the Big Bang. This plasma contains roughly equal numbers of quarks and antiquarks. As a result of the high energy density of the QGP phase, many strange-antistrange (ss) quark pairs 1 are liberated from the quantum vacuum. The plasma cools and transitions into a hadron gas, producing nucleons, hyperons, mesons, and their antiparticles.Nucleons (protons and neutrons) contain only up and down valence quarks, while hyperons (Λ, Σ, Ξ, Ω) contain at least one strange quark in its 3-quark valence set. A hypernucleus is a nucleus that contains at least one hyperon in addition to nucleons. All hyperons are unstable, even when bound in nuclei. The lightest bound hypernucleus is the hypertriton ( 3 Λ H), which consists of a Λ hyperon, a proton, and a neutron. The first observation of any hypernucleus was made in 1952 using a nuclear emulsion cosmic ray detector (5). Here, we present the observation of an antimatter hypernucleus. Production of antinuclei:Models of heavy-ion collisions have had good success in explaining the production of nuclei by assuming that a statistical coalescence mechanism is in effect during the late stage of the collision evolution (4, 6). Antinuclei can be produced through the same coalescence mechanism, and are predicted to be present in cosmic rays. An observed high yield could be interpreted as an indirect signature of new physics, such as Dark Matter (7, 8). Heavy-ion collisions at RHIC provide an opportunity for the discovery and study of many antinuclei and antihypernuclei.The ability to produce antihypernuclei allows the study of all populated regions in the 3-dimensional chart of the nuclides. The conventional 2-dimensional chart of the nuclides organizes nuclear isotopes in the (N, Z) plane, where N is the number of neutrons and the Z is the number of protons in the nucleus. This chart can be extended to the negative sector in the (N, Z) plane by including antimatter nuclei. Hypernuclei bring a third dimension into play, based on the strangeness quantum number of the nucleus. The present study probes the territory of antinuclei with non-zero strangeness ( Fig. 1), where proposed ideas (9-12) related to t...

We present the first measurements of identified hadron production, azimuthal anisotropy, and pion interferometry from Au + Au collisions below the nominal injection energy at the BNL Relativistic Heavy-Ion Collider (RHIC) facility. The data were collected using the large acceptance solenoidal tracker at RHIC (STAR) detector at √ s NN = 9.2 GeV from a test run of the collider in the year 2008. Midrapidity results on multiplicity density dN/dy in rapidity y, average transverse momentum p T , particle ratios, elliptic flow, and Hanbury-Brown-Twiss (HBT) radii are consistent with the corresponding results at similar √ s NN from fixed-target experiments. Directed flow measurements are presented for both midrapidity and forward-rapidity regions. Furthermore the collision centrality dependence of identified particle dN/dy, p T , and particle ratios are discussed. These results also demonstrate that the capabilities of the STAR detector, although optimized for √ s NN = 200 GeV, are suitable for the proposed QCD critical-point search and exploration of the QCD phase diagram at RHIC.