- Seminars at CIPS

The National Transport Code Collaboration seeks to develop a code for studying anomalous transport in toroidally confined plasma. Much as experimentalists take diagnositics to facilities to study new phenomena experimentally, the NTCC Demonstration Code acts as a computational facility for new models for transport. Theorists must provide their code in the form of an object satisfying certain interface constraints. Having done so, they are in an environment where they may, on-line, check their theory against experiment and run their models with a web-invocable, graphical user interface. The next IAEA meeting will showcase results from the NTCC Demonstration Code.

The Department of Pediatrics at the University of Colorado School of Medicine is developing a bank of video cases that correspond to the Pediatric National Curriculum. The cases are used in a WWW/CD-ROM hybrid program with "virtual" Problem-Based Learning (PBL) groups that conduct computer-mediated case discussions between a faculty mentor and students at multiple clinical sites. To the degree possible, our cases simulate a real encounter by using computer-based digital video rather than text, forcing the student to glean the information from the patient encounter and develop the visual recognition skills that are important in pediatric medicine. Unlike most adult patients that can verbally provide a history, pediatric patients very often cannot. Physicians must recognize visual and auditory cues to accurately diagnose a child. Digital video cases also afford the opportunity to model appropriate professional behavior and communication skills. These opportunities are especially important for difficult situations in which the physician needs effective communication skills. Students are often not allowed in some of those interviews and when they are included the variability of the modeling is problematic. Instructional strategies integrate group work, problem solving, and mentoring to support this teaching method. Collaborative learning across the Internet, combined with digital video cases has the potential to have a tremendous impact not only on medical education, but also on distance learning in general.

The invention of the ion trap has led to impressive technological advances in physical measurements as well as the discovery of new states of matter. The motion of ions in the axisymmetric Paul trap has been successfully described by a classical Hamiltonian model, in which the ions move under the combined influence of the focusing RF quadrupole field and their mutual Coulomb repulsion. However, most current experiments are carried out in an asymmetric mode, in which the time-averaged "pseudopotential" is a triaxial ellipsoid. Critical points of this potential yield equilibria of ion "crystals", which "melt" to form "clouds". Chaotic motion can lead to enhanced RF heating and loss of ions. We present a classical description of one and two-ion dynamics in an elliptic trap. Stability boundaries for single ion motion are shown to decreased rapidly with increasing asymmetry. The two-ion motion takes place in a fully three-dimensional pseudopotential possessing three isolated critical points known as Morse saddles. Stability is lost when a pair of critical points change type. The results show that global confinement can prevail in the presence of unstable local equilibria. Chaotic motion is studied using four-dimensional surfaces of section.

Many data analysis problems can be approached by fitting a probability distribution to the data, but there are few models for distributions of images. By a distribution of images we mean a distribution describing images resulting from some process, for example sub-images selected from mammograms by a computer-aided diagnosis system. Previous attempts to model image probabilities either model the probability of some features of the image or are only suited for modeling images of textures; they seem unsuitable for modeling images of more structured objects. To address this problem we formulate a set of models for probability distributions on image spaces, which we call Hierarchical Image Probability or HIP models. They are hierarchical because they explicitly represent image structure at several length scales, and finer scale structures are conditioned on coarser scales. To make the model tractable we factor the distribution over scale and position. Such factoring would make it impossible to capture long-range correlations that arise from the objects being imaged. To fix this we introduce a further hierarchy of hidden variables whose probabilities also factor over scale and position. Since they are unknown, they must be summed over or marginalized to evaluate the image probability, and the summation reintroduces long-range correlations. We present algorithms for performing this sum and for finding the model parameters with maximum likelihood estimation. We have obtained encouraging preliminary results on the problems of detecting various objects in SAR images, target recognition in optical aerial images, and mass detection in mammograms.

This talk demonstrates numerous computer systems that teach physics, chemistry, mathematics, medicine or engineering. These systems have resulted in increased student time on task, improved student grades and reduced faculty costs and are used by more than 4,000 students at over 20 universities.

In physics, we have built an electronic homework system which results in improved test and final exam scores, sometimes adding a full letter grade to the student score. Additionally, the weaker students receive the greatest benefit from this system. In chemistry, Web-based homework program includes thirty-five interactive discovery environments that provide guided inquiry, feedback to student responses and tracking of student performance. Intelligent tutors in chemistry provide levels of customized responses to expose students to the depths of chemical reactions. One tutor enables students to directly manipulate images and work with a palette of tools for placing and moving symbols. Tutors have been tested with over 900 students and show positive improvement in final exams. The electronic homework system provides an open architecture allowing for rapid extensions to new departments.

In addition, engineering tutors provide animated 3D tooling solutions of student designs and advice about relative costs. Evaluation demonstrates that these tutors are as effective as several lectures and homework assignments within a traditional classroom setting. A microbiology tutor provides visual support for an entire undergraduate course in molecular biology with rich 3-D animations depicting production of proteins through the interaction of DNA and RNA. A mathematics tutor uses machine learning to individualize problems and hints.

Each demonstrated project has been evaluated for effectiveness and efficiency. The talk will show how technology's impact on leaning has been quantified and that these systems can be of general use on a national scale.

Results are presented of ALEGRA simulations of the x-ray pulse shape from shot 26 of Sandia's Z machine. ALEGRA is Sandia's multi-dimensional, arbitrary Lagrangian-Eulerian MHD code. Shot 26 produced 180 TW of x-ray power in a 7.5-ns FWHM pulse. This shot was chosen because other MHD codes (MACH II and Darrell Peterson's code from LANL) also have simulated shot 26, thereby providing the opportunity to compare ALEGRA to other codes as well as to data. Discussed in this talk are the effects on x-ray pulse shape of: (i) true void versus plasma fill inside the liner, (ii) differing interface tracking schemes, and (iii) differing levels and models of density perturbations.

Particle acceleration occuring at or near the time solar flares is common, but it is not so common for high energy particles to reach the Earth's space environment. Solar energetic particles can have very destructive consequences for satellites and astronauts in space; thus, an import question to ask is if a large, destructive proton event will occur in conjuction with a particular solar flare or an associated coronal mass ejection. Acceleration of high energy electrons in flares is generally observed via hard X-ray (bremsstrahlung) and microwave (gyrosychriotron) emissions. In two earlier studies, Kiplinger has found a very high association between an uncommon hard X-ray signature called "progresssive spectral hardening" and interplanetary proton events. Those studies have been been extended from ~330 to more than 700 solar flares observed in hard X-rays by the Solar Maximum Mission (SMM). The results confirm the robustness for the association of progressive spectral hardening with major interplatetary proton events. Some episodes of progressive hardening are also associated with high energy neutrons seen at Earth. Gamma ray lines in flares also result from high energy ion production. A new, ongoing study of all gamma ray line flares seen by SMM compares times of emission of gamma ray lines with occurrences of progressive spectral hardening. These results and their implications will be discussed.

We have examined the photoelectric charging of dust 90-106 microns in diameter dropped through UV illumination and dropped past a UV illuminated surface having a photoelectron sheath. Experiments are performed in vacuum with illumination from a 1 kW Hg-Xe arc lamp that has a spectrum extending to 200 nm (6.2 eV). We present and compare the photoelectric charging properties of particles composed of zinc, copper, graphite, lunar regolith simulant (JSC-1), and martian regolith simulant (JSC Mars-1). We find that the photoelectric charging properties of the elemental materials are consistent with charging models calculated from the theoretical capacitance and charge on an isolated spherical grain. Dust dropped through UV illumination loses electrons due to photoemission, while dust dropped past an illuminated surface gains electrons from the photoelectron sheath. The photoelectric charging properties of JSC-1 and JSC Mars-1 are more difficult to interpret due to residual charge on the dust. The results suggest that JSC Mars-1 is more susceptible to photoelectric charging than JSC-1. The relation of this work to similar phenomena in the solar system is discussed.

The period doubling bifurcation process in the two-dimensional area preserving mapping is investigated on the basis of simmetry structure analysis. In particular a case of the period-4 orbits in the standard map has been studied throughly to analyze boundary islands formation around the principal period-4 island, and the onset of the hyperbolic bifurcation without reflection. It is illustrated explicitly that the hyperbolic bifurcation without reflection gives rise to the birth of twin orbits with the periodicity of the mother orbit.

The effect of poloidally mode coupled, ballooning type electrostatic drift waves on a magnetic island has been studied both analytically and numerically. It has been shown quantitatively that particle orbits become stochastic and their behavior can be a possible candidate for the radial plasma transport across a magnetic island of a tokamak. The transport is significant in that it takes place even when the flux surface is not destroyed. The mechanism of the stochasticity generation is understood as an overlapping of secondary islands caused by resonance between periodic particle motions in the magnetic island and Fourier modes of E x B drift due to the electrostatic drift waves. The diffusion process perpendicular to the island magnetic surface has been shown to follow the Gaussian type and can be influential for the deterioration of the plasma confinement. In addition, local diffusion process in the vicinity of Kolmogorov, Arnold and Moser (KAM) surfaces is discussed.

Finite-length equilibria occur in a number of intense-beam and plasma applications. Penning traps permit the study of intra-beam collective effects, as the additional freedom gained from having an internal conductor permits greater control over the plasma profile, so that monotonic, but not constant, plasma profiles can be obtained. On the basis that the thermal velocity of background neutrals and the drift velocity of the electrons are much lower than the thermal velocity of the electrons, and the rotation frequency is small compared to the gyrofrequency, the equilibrium equation can be reduced to a self-consistent Poisson equation where the source depends on the potential. We solve for these equilibria using a Gauss-Seidel relaxation method. Our results show the shape of the equilibria for various electrode configurations.

In 1985 it was shown that linear analysis techniques were inadequate to describe the behavior of the magnetosphere, and so nonlinear studies of the magnetosphere were begun. When a finite dimension for an attractor was found in magnetospheric data it was thought that conclusive proof was found for chaotic behavior. However, later studies found that random noise with a certain frequency spectrum, called colored random or pink noise, could also give a finite result for the correlation dimension. Since then a controversy has been raging as to whether the magnetospheric was chaotic or stochastic. The definition and method for computing the correlation dimension will be discussed as well as the properties of colored random noise. Also to be discussed is how the properties of colored random noise can lead to a test to determine the difference between a colored random noise time series and a time series from ordinary differential equations that are chaotic.

We study the global stability of charged dust grains orbiting an axisymmetric planet with co-rotating magnetic field. The magnetic field and induced electric field are described in an inertial frame using the magnetic stream function $\Psi$. The combined gravitational, magnetic, and electric forces are modelled by a two dimensional effective potential $U^e(\rho,z)$, parametrized by the conserved angular momentum $\pp$. The critical points of $U^e$ then locate the equilibrium circular orbits, nonequatorial as well as equatorial. The stable equilibria form the nuclei of potential wells, which can contain large populations of dust grains. These potential wells have their own topological structure, so that a particle which loses local stability can still be trapped globally. Explicit Lyapunov stability boundaries are derived for both positive and negative charges in both prograde and retrograde orbits. Thus, radial stability is lost when a critical point of $U^e$ undergoes a ta! ngent bifurcation, while transverse stability is lost via a pitchfork bifurcation. For a given position near a given planet stability depends only on the charge-to-mass ratio $q/m$, which for a spherical dust grain is proportion to $\Phi/a^2$, where $\Phi$ is the ambient plasma potential and $a$ is the grain radius. The results are applied to Saturn and Jupiter.