Hertz Foundation

Quantitative evidence is presented that counter the normative cultural ones as to who does physics and why one does physics, helps to support women and minoritized groups in envisioning a future for themselves in physics.

Professor Edward Teller (1908–2003) was a remarkable individual who was shaped by and participated in the events of the most dramatic century in human history. He actively participated in those events through his intellectual contributions as a student, an educator, as an applied physicist, and through his technical leadership of the United States’ nuclear defense programs. One of his enduring contributions has been to demonstrate the power of thinking deeply about important problems and to deal directly with them–no matter what the consequences. This paper follows the outline of my talk given at the Teller Centennial, which took place in Livermore on May 28. The talk is

We consider a channel where the sender has access to channel state information and wishes to send both the pure information and a description of the channel state to the receiver. An achievable tradeoff region between pure information rate and state estimation error is proposed for a discrete memoryless channel with an arbitrary state distortion measure.

We have previously found in the Hubble Deep Field (HDF) a significant angular correlation of faint, high-color redshift objects on scales below 1'', or several kiloparsecs in metric size. A correlation at this scale is most likely due to physical associations. We examine the correlation and nearest neighbor statistics to conclude that 38% of these objects in the HDF have a companion within 1'' (or ~6 kpc), 3 times the number expected in a random distribution with the same number of objects; the total approaches 1.5 objects by separations of 10''. We next examine three possible dynamical scenarios for these object multiplets: (1) the objects are star-forming regions within normal galaxies, whose disks have been relatively dimmed by K-correction and surface brightness dimming; (2) they are fragments merging into large galaxies; (3) they are satellites accreting onto parent normal L* galaxies. We find that hypothesis 1 is most tenable. First, large galaxies in the process of a merger formation would have accumulated too much mass in their centers (5 × 1012 M☉ inside 2 kpc) to correspond to any abundant category of present-day objects. Second, accretion by dynamical friction occurs with a predictable slope in density versus radius that is not seen among the faint HDF objects. Since the dynamical friction time is roughly 1 Gyr, a steady state should have been reached by redshift z 5. In the context of these two dynamical scenarios, we consider the possible effects of a gradient in mass-to-light ratio caused by induced star formation during infall. We note that star-forming regions within galaxies clearly present no dynamical problems, but also that large spirals would still appear as such in the HDF, which leads us to favor a scenario in which the faint compact sources in the HDF are giant star-forming regions within small normal galaxies, such as Magellanic irregulars. Last, we note that the excess number of correlated objects near a given faint source approaches 1.5, suggesting that the previous counts of objects have overestimated the number of galaxies by a factor of 2.5 while underestimating their individual luminosities by the same factor.

This paper extends the quasilinear theory of Kulsrud & Anderson to assess the effects of realistically long velocity correlation times on the turbulent amplification of a very weak magnetic field. A computer simulation is presented that tracks the growth of the magnetic energy in a turbulent plasma at a single point moving with the turbulent flow. The velocities are assumed to conform to the ideas of Kraichnan concerning Lagrangian correlation times, and are modeled as a set of randomly generated pulses chosen to reproduce the correct two-time Lagrangian correlation tensor. The model is simple computationally and can be used to calculate the growth rate of the magnetic energy for arbitrarily high magnetic Reynolds numbers. The simulations show that the magnetic energy grows roughly half as fast as predicted in the short correlation time approximation of Kulsrud & Anderson's quasilinear theory. In a separate analysis, the effects of nonzero correlation times are considered using an analytic method developed by van Kampen. The growth rate is expanded, roughly speaking, in powers of the correlation time divided by the time required for the energy to exponentiate once. The first two terms in the series are calculated. In themselves, these two terms do not exactly determine the growth rate, but they are consistent with the numerical results. The analytic treatment is included mostly for completeness and because it offers some physical understanding of the problem. The main conclusion of the paper is that velocity correlation times do not play an important role in the growth of the magnetic energy. As a result, Kulsrud & Anderson's short correlation time analysis of the spectrum of amplified small-scale fields should be approximately correct.

Photometric monitoring of the gravitational lens system 0957+561A,B in the g\nand r bands with the Apache Point Observatory (APO) 3.5 m telescope during 1996\nshows a sharp g band event in the trailing (B) image light curve at the precise\ntime predicted from the observation of an event during 1995 in the leading (A)\nimage with a delay of 415 days. This success confirms the "short delay," and\nthe lack of any feature at a delay near 540 days rejects the "long delay" for\nthis system, resolving a long-standing controversy. A series of statistical\nanalyses of our light curve data yield a best fit delay of 417 +/- 3 days (95%\nconfidence interval). Recent improvements in the modeling of the lens system\n(consisting of a galaxy and cluster) allow us to derive a value of the global\n(at z = 0.36) value of Hubble's constant H_0 using Refsdal's method, a simple\nand direct distance determination based on securely understood physics and\ngeometry. The result is H_0 = 63 +/- 12 km/s/Mpc (for Omega = 1) where this 95%\nconfidence interval is dominated by remaining lens model uncertainties.\n

Several researchers have considered the analysis of control system reliability for systems without plant/model mismatch. In this manuscript we review several of these proposed analysis tools, including some of the strongest, and show that the results are either conservative, computationally expensive, or incorrect. We then develop necessary and sufficient conditions for many forms of reliability defined in the literature, and since the model is always an imperfect representation for the true process, we extend the resulting analysis tools to uncertain systems.

A megachannel pulse-height analysis system using a PDP-8/E computer and two moving-head disk memories has been developed. The system has a storage capacity of 220 memory locations, is capable of processing 1100 events/s, and provides on-line sorting and disk storage. An X- or Y-pulse-height spectrum in coincidence with one or several arbitrary pulse-height windows can be assembled in core for scope display and spectral analysis within 2 to 20 seconds. Reconstruction of a complete X- or Y-pulse-height spectrum requires about 3 minutes.