Search

Learning, especially rapid learning, is critical for survival. However, learning is hard; a large number of synaptic weights must be set based on noisy, often ambiguous, sensory information. In such a high-noise regime, keeping track of probability distributions over weights is the optimal strategy. Here we hypothesize that synapses take that strategy; in essence, when they estimate weights, they include error bars. They then use that uncertainty to adjust their learning rates, with more uncertain weights having higher learning rates. We also make a second, independent, hypothesis: synapses communicate their uncertainty by linking it to variability in postsynaptic potential size, with more uncertainty leading to more variability. These two hypotheses cast synaptic plasticity as a problem of Bayesian inference, and thus provide a normative view of learning. They generalize known learning rules, offer an explanation for the large variability in the size of postsynaptic potentials and make falsifiable experimental predictions.

The dominant view in neuroscience is that changes in synaptic weights underlie learning. It is unclear, however, how the brain is able to determine which synapses should change, and by how much. This uncertainty stands in sharp contrast to deep learning, where changes in weights are explicitly engineered to optimize performance. However, the main tool for that, backpropagation, has two problems. One is neuro-science related: it is not biologically plausible. The other is inherent: networks trained with this rule tend to forget old tasks when learning new ones. Here we introduce the Dendritic Gated Network (DGN), a variant of the Gated Linear Network, which offers a biologically plausible alternative to backpropagation. DGNs combine dendritic ‘gating’ (whereby interneurons target dendrites to shape neuronal responses) with local learning rules to yield provably efficient performance. They are significantly more data efficient than conventional artificial networks, and are highly resistant to forgetting. Consequently, they perform well on a variety of tasks, in some cases better than backpropagation. Importantly, DGNs have structural and functional similarities to the cerebellum, a link that we strengthen by usingin vivotwo-photon calcium imaging to show that single interneurons suppress activity in individual dendritic branches of Purkinje cells, a key feature of the model. Thus, DGNs leverage targeted dendritic inhibition and local learning – two features ubiquitous in the brain – to achieve fast and efficient learning.

Sensory processing is hard because the variables of interest are encoded in spike trains in a relatively complex way. A major goal in studies of sensory processing is to understand how the brain extracts those variables. Here we revisit a common encoding model in which variables are encoded linearly. Although there are typically more variables than neurons, this problem is still solvable because only a small number of variables appear at any one time (sparse prior). However, previous solutions require all-to-all connectivity, inconsistent with the sparse connectivity seen in the brain. Here we propose an algorithm that provably reaches the MAP (maximum a posteriori) inference solution, but does so using sparse connectivity. Our algorithm is inspired by the circuit of the mouse olfactory bulb, but our approach is general enough to apply to other modalities. In addition, it should be possible to extend it to nonlinear encoding models.SummarySensory systems must infer latent variables from noisy and ambiguous input. MAP inference – choosing the most likely latent variable given the sensory input – is one of the simplest methods for doing that, but its neural implementation often requires all-to-all connectivity between the neurons involved. In common sensory contexts this can require a single neuron to connect to hundreds of thousands of others, which is biologically implausible. In this work we take inspiration from the ‘sister’ mitral cells of the olfactory system – groups of neurons associated with the same input channel – to derive a method for performing MAP inference using sparse connectivity. We do so by assigning sister cells to random subsets of the latent variables and using additional cells to ensure that sisters correctly share information. We then derive the circuitry and dynamics required for the sister cells to compute the original MAP inference solution. Our work yields a biologically plausible circuit that provably solves the MAP inference problem and provides experimentally testable predictions. While inspired by the olfactory system, our method is quite general, and is likely to apply to other sensory modalities.

The two basic processes underlying perceptual decisions – how neural responses encode stimuli, and how they inform behavioral choices – have mainly been studied separately. Thus, although many spatiotemporal features of neural population activity, or “neural codes,” have been shown to carry sensory information, it is often unknown whether the brain uses these features for perception. To address this issue, we propose a new framework centered on redefining the neural code as the neural features that carry sensory information used by the animal to drive appropriate behavior; that is, the features that have an intersection between sensory and choice information. We show how this framework leads to a new statistical analysis of neural activity recorded during behavior that can identify such neural codes, and we discuss how to combine intersection-based analysis of neural recordings with intervention on neural activity to determine definitively whether specific neural activity features are involved in a task.

If the brain processes incoming data efficiently, information should degrade little between early and later neural processing stages, and so information in early stages should match behavioral performance. For instance, if there is enough information in a visual cortical area to determine the orientation of a grating to within 1 degree, and the code is simple enough to be read out by downstream circuits, then animals should be able to achieve that performance behaviourally. Despite over 30 years of research, it is still not known how efficient the brain is. For tasks involving a large number of neurons, the amount of information encoded by neural circuits is limited by differential correlations. Therefore, determining how much information is encoded requires quantifying the strength of differential correlations. Detecting them, however, is difficult. We report here a new method, which requires on the order of 100s of neurons and trials. This method relies on computing the alignment of the neural stimulus encoding direction, f′, with the eigenvectors of the noise covariance matrix, Σ. In the presence of strong differential correlations, f′ must be spanned by a small number of the eigenvectors with largest eigenvalues. Using simulations with a leaky-integrate-and-fire neuron model of the LGN-V1 circuit, we confirmed that this method can indeed detect differential correlations consistent with those that would limit orientation discrimination thresholds to 0.5-3 degrees. We applied this technique to V1 recordings in awake monkeys and found signatures of differential correlations, consistent with a discrimination threshold of 0.47-1.20 degrees, which is not far from typical discrimination thresholds (1-2 deg). These results suggest that, at least in macaque monkeys, V1 contains about as much information as is seen in behaviour, implying that downstream circuits are efficient at extracting the information available in V1.

Systems neuroscience seeks explanations for how the brain implements a wide variety of perceptual, cognitive and motor tasks. Conversely, artificial intelligence attempts to design computational systems based on the tasks they will have to solve. In the case of artificial neural networks, the three components specified by design are the objective functions, the learning rules, and architectures. With the growing success of deep learning, which utilizes brain-inspired architectures, these three designed components have increasingly become central to how we model, engineer and optimize complex artificial learning systems. Here we argue that a greater focus on these components would also benefit systems neuroscience. We give examples of how this optimization-based framework can drive theoretical and experimental progress in neuroscience. We contend that this principled perspective on systems neuroscience will help to generate more rapid progress.

Most models of decision-making suggest that confidence, the ‘feeling of knowing’ that accompanies our choices, is constructed as the decision unfolds. However, more recent studies have noted that processes occurring after we commit to a particular choice also affect this subjective belief. This leads to the following question: when are we better judges of ourselves? If, after a decision, evidence continues to accumulate in an unbiased manner, then our confidence judgements should improve. Conversely, if post-decisional information processing is biased, our sense of confidence could be distorted, and so our confidence judgements should degrade with time. We briefly discuss recently proposed models of post-decisional evidence accumulation, and explore whether, and how, biases in confidence could arise.

A fundamental problem in neuroscience is understanding how working memory -- the ability to store information at intermediate timescales, like 10s of seconds -- is implemented in realistic neuronal networks. The most likely candidate mechanism is the attractor network, and a great deal of effort has gone toward investigating it theoretically. Yet, despite almost a quarter century of intense work, attractor networks are not fully understood. In particular, there are still two unanswered questions. First, how is it that attractor networks exhibit irregular firing, as is observed experimentally during working memory tasks? And second, how many memories can be stored under biologically realistic conditions? Here we answer both questions by studying an attractor neural network in which inhibition and excitation balance each other. Using mean field analysis, we derive a three-variable description of attractor networks. From this description it follows that irregular firing can exist only if the number of neurons involved in a memory is large. The same mean field analysis also shows that the number of memories that can be stored in a network scales with the number of excitatory connections, a result that has been suggested for simple models but never shown for realistic ones. Both of these predictions are verified using simulations with large networks of spiking neurons., Accepted for publications in PLoS Comp. Biol

We calculate the firing rate of the quadratic integrate-and-fire neuron in response to a colored noise input current. Such an input current is a good approximation to the noise due to the random bombardment of spikes, with the correlation time of the noise corresponding to the decay time of the synapses. The key parameter that determines the firing rate is the ratio of the correlation time of the colored noise, τs, to the neuronal time constant, τm. We calculate the firing rate exactly in two limits: when the ratio, τs/τm, goes to zero (white noise) and when it goes to infinity. The correction to the short correlation time limit is O(τs/τm), which is qualitatively different from that of the leaky integrate-and-fire neuron, where the correction is O(√τs/τm). The difference is due to the different boundary conditions of the probability density function of the membrane potential of the neuron at firing threshold. The correction to the long correlation time limit is O(τm/τs). By combining the short and long correlation time limits, we derive an expression that provides a good approximation to the firing rate over the whole range of τs/τm in the suprathreshold regime—that is, in a regime in which the average current is sufficient to make the cell fire. In the subthreshold regime, the expression breaks down somewhat when τs becomes large compared to τm.

The experimental demonstration of a four cavity W-band (93 GHz) gyroklystron amplifier is reported. The gyroklystron has produced 67 kW peak output power and 28% efficiency in the TE011 mode using a 55 kV, 4.3 A electron beam. The full width at half maximum instantaneous bandwidth is greater than 460 MHz, a significant increase over the bandwidth demonstrated in previous W-band gyroklystron amplifier experiments. The amplifier is unconditionally stable at this operating point. Experimental results are in good agreement with theoretical predictions.

The brain is easily able to process and categorize complex time-varying signals. For example, the two sentences, “It is cold in London this time of year” and “It is hot in London this time of year,” have different meanings, even though the words hot and cold appear several seconds before the ends of the two sentences. Any network that can tell these sentences apart must therefore have a long temporal memory. In other words, the current state of the network must depend on events that happened several seconds ago. This is a difficult task, as neurons are dominated by relatively short time constants—tens to hundreds of milliseconds. Nevertheless, it was recently proposed that randomly connected networks could exhibit the long memories necessary for complex temporal processing. This is an attractive idea, both for its simplicity and because little tuning of recurrent synaptic weights is required. However, we show that when connectivity is high, as it is in the mammalian brain, randomly connected networks cannot exhibit temporal memory much longer than the time constants of their constituent neurons.

We develop a framework for generating oscillations in ring attractor networks with firing rate adaptation. We show the relationship between the frequency of rotation around the ring of the shifting bump of activity, the adaptation variable and other model parameters using perturbation theory. The analytic solutions are validated against simulations of such networks. Further preliminary findings indicate that the frequency of these networks can be simply controlled using an external stimulus. The mechanism developed here could potentially be used for temporal coding of position through interference of oscillators of different frequencies.

We report the first experimental operation of overmoded first and second harmonic gyrotwystron amplifier configurations. Both devices utilize a single cavity which is driven near 9.87 GHz in the TE011 mode, heavily attenuated drift tubes, and long tapered output waveguide sections. A magnetron injection gun produces a 460 kV, 245 A beam with a maximum average perpendicular‐to‐parallel velocity ratio approximately equal to one. The axial magnetic field profile is sharply tapered in the output section. Peak powers above 21 MW are achieved in 1 μs pulses with an efficiency exceeding 22% and a large signal gain near 24 dB in the first harmonic tube. The second harmonic tube achieves nearly 12 MW of the peak power with an efficiency of 11% and a gain above 21 dB. First harmonic amplifier performance is limited principally by competition from a fundamental mode output waveguide interaction; the second harmonic tube is limited by both travelling wave output modes and by a down‐taper oscillation.

Phase-locking in a 34.5-GHz special complex cavity gyrotron oscillator operating at the second harmonic of the electron cyclotron frequency was studied. Injection of the locking power was made via a quasi-optical circulator connected to the gyrotron output. Locking bandwidth was measured by comparing the phase of the injection signal and output signal using a balanced mixer. Locking was observed with input power level as low as 40 dB below the gyrotron output power. The locking bandwidth is, however, narrower than in gyrotrons operating at the fundamental cyclotron frequency which may be attributed to the longer resonant cavity in the second harmonic gyrotron and the corresponding larger value of external quality factor. The measurements are roughly in agreement with predictions of Adler's phase-locking equation which is given for our system in terms of powers propagating in the output waveguide toward and away from the gyrotron cavity. >

The performance of a two‐cavity second harmonic gyroklystron employing a short, mixed‐mode output cavity with stepped radial transitions is described. The short cavity was employed to prevent low frequency spurious oscillations that can compete with the harmonic signal. The radial rf field profile in the cavity and the output radiation consisted of both the TE01 and TE02 circular waveguide modes. This device produced a peak output power of 20 MW at 19.782 GHz, with an efficiency of 23% and a gain of 26 dB. The nominal mode mixture of the radiated power during optimal operation was found to be about 60% TE02 and 40% TE01. Variations from this ratio are observed at lower powers and are attributed to shifts in the cavity rf field profiles. Systematic studies of amplifier performance as a function of beam parameters and magnetic field profile are described. The sensitivities of output power production to changes in operating parameters are compared to results from earlier harmonic and fundamental gyroklystrons. The present device is found to be more stable to parasitic oscillations and less sensitive to variations in beam current and output cavity magnetic field than previous harmonic gyroklystrons that employed longer, smoothly transitioned cavities. However, both the peak power and efficiency in the present device are lower than the 30 MW at 28% efficiency obtained with the final smooth‐cavity tube. The tradeoffs between the two output cavity concepts will be discussed.

The theory of phase-locked gyrotrons operating at cyclotron harmonics is developed, with particular attention given to the effects of electron velocity spread. Extensive studies are made for operation at the second harmonic in which phase locking is achieved by subharmonic prebunching. One result of these studies is that we are able to predict, for a range of velocity spreads, the parameters that yield the highest efficiency. More important, we find that for electron beams with velocity spread and a finite transit angle in the drift region, the bandwidth of stable operation is drastically reduced. This effect can be mitigated by properly tapering the external magnetic field there. Such tapering is necessary in devices where the bandwidth is required to be a reasonable fraction of 1/Q. >

For gyrodevices with a given operating frequency and a fixed value of the external magnetic field two kinds of operation are compared: Doppler frequency up-shifted operation at the fundamental cyclotron resonance and cyclotron harmonic operation. When the electron velocity spread is negligibly small the maximum efficiency of the Doppler up-shifted operation exceeds that of harmonic operation. However, as the velocity spread grows the Doppler up-shifted efficiency degrades faster. The relevant quantity when deciding between fundamental, Doppler up-shifted, and harmonic operation turns out to be the velocity spread divided by the square root of the beam current. If this quantity is small, fundamental, Doppler upshifted operation is more efficient than harmonic operation; if it is large harmonic operation is more efficient. The efficiency dependence of both kinds of operation on electron velocity spread is studied. A comparative analysis of designs relevant to the planned University of Maryland experiments with relativistic gyro-amplifiers is done. >

We report the operating characteristics of a sequence of two-cavity second harmonic gyroklystrons which are derived in part from a previous fundamental tube and utilize output cavities which resonate at twice the drive frequency. We present results from the design simulations as well as details of the stable range of operating parameters. While the harmonic tube is somewhat more susceptible to spurious oscillations and more sensitive to parameter variations than the fundamental device, there is still considerable parameter space available for amplifier operation. Peak powers above 30 MW are obtained with efficiencies greater than 28% and large signal gains of 27 dB. These results depend critically on the magnetic field profile which has a slight up-taper at the optimum operating point. The nominal beam parameters include a pulse length of 1 /spl mu/s, a voltage near 450 kV, a current in the range 235-245 A, and a perpendicular to parallel velocity ratio (/spl alpha/=v/sub /spl perp///v/sub z/) near one. >

The design of a gyroklystron experiment capable of fulfilling the expected requirements of future electron-positron colliders is presented. The design is an extension of previous work that utilized the interaction of a 500 kV beam and TE/sub 01//TE/sub 02/ circular cavities. Details of an electron gun design which is capable of producing an 800 A, 1 /spl mu/s beam are given. Modifications to an existing code which enhance the system modeling are described before the circuit simulation results are summarized. >

For imaging radar and for satellite and space communication (e.g. NASA's deep space network), it is important that the bandwidth be as large as possible. Here we derive a formalism for computing the phase locking bandwidth that can be achieved in a gyrotron oscillator while varying the beam voltage. As an example, a second harmonic TE02/03 gyrotron is considered. For this device, the effective bandwidth can be increased by a factor of about 3 compared with the fixed voltage case by allowing the beam voltage to change together with the input locking signal.

Experimental results for a 10 GHz TE/sub 01/ mode three-cavity gyroklystron with a tunable penultimate cavity are presented. The electron beam was produced by a pulse line modulator and a magnetron injection gun which operates to 433 kV and 225 A with 1 mu s flat-top. Three-cavity circuits have produced a peak power of 27 MW with efficiency of 32% and pulse energy of 39 J. A maximum gain of 50 dB was achieved at a peak power of 20 MW, and a maximum efficiency of 37% was achieved at a peak power of 16 MW. >

We investigate the phenomenon of transverse mode interference (TMI) in electromagnetic waveguide structures. We find that when waveguides are connected to a structure with sufficiently abrupt discontinuities to produce substantial mode conversion, 100% reflection at a finite set of frequencies is generic. At a frequency where 100% reflection occurs, degeneracies are found in the associated closed cavity configuration (end walls on the waveguides). The signatures of these degeneracies are mode crossings, which in turn are associated with resonances of the TMI structure.

The scattering matrix formulation for complex right-circular activities is extended to coaxial circuits with variable inner radii. The modified eigenvectors, which include the TEM wave, and the modified boundary conditions are presented. The properties of several configurations are examined and transmission measurements are shown to be in good agreement with theory. >

The performance of four overmoded (TE011 circular), two‐cavity gyroklystron amplifier tubes operating in X band is reported. A summary of the theoretical design procedure is presented, followed by a description of the experimental setup. The stability and amplification properties of the four gyroklystron tubes are subsequently described, each being modified in sequence to improve performance. The fourth tube produced 2.0–2.7 MW pulses at 9.87 GHz for 0.5–1.0 μs, and exhibited gains of 17–19 dB and efficiencies of 5%. The results were obtained using electron beams with voltages, currents, and computed beam alphas (v⊥/vz) of 407–425 kV, 115–135 A, and 0.8–1.0, respectively. The presence of instabilities in the gun downtaper and drift tube prevented operation at the design point (500 kV, 160 A, and alpha 1.5). The effects of varying input frequency, input power, and beam parameters (voltage, current, magnetic compression) on the amplifier operation are reported. Detailed descriptions of the instabilities obs...

Experimental studies of amplification in a two-cavity X-band gyroklystron are reported. The system utilizes a thermionic magnetron injection gun at voltages up to 440 kV and currents up to 190 A in 1-μs pulses. Optimum performance is achieved by tapering the magnetic-field profile. Peak powers of 20 MW in the TE01 mode at 9.87 GHz are measured with calibrated crystals and with methanol calorimetry. Resultant efficiencies are in excess of 31% and large-signal gains surpass 26 dB. The experimental results are in good agreement with simulated results from a partially self-consistent, nonlinear, steady-state code.

A high-average-power FEL amplifier operating at submillimeter frequencies is under development at the University of Maryland. Program goals are to produce a cw, ∼1 MW, FEL amplifier source at frequencies between 280 and 560 GHz. To this end, a high-gain, high-efficiency, tapered FEL amplifier using a sheet electron beam and a short-period (superconducting) wiggler has been chosen. Development of this amplifier is progressing in three stages: 1) beam propagation through a long length (∼ 1 m) of a short-period ( λ w = 1 cm) wiggler, 2) demonstration of a proof-of-principle amplifier experiment at 98 GHz, and 3) designs of a superconducting tapered FEL amplifier meeting the ultimate design goal specifications.

Expressions are derived for RF phase stability in conventional klystrons and gyroklystrons. Phase noise is found to depend on electron energy in a completely different way for the two types of devices, due to the inherent differences in gun dynamics and interaction mechanisms. In general, phase stability is better in gyroklystrons operating at voltage >

SummaryWhen making a decision, one must first accumulate evidence, often over time, and then select the appropriate action. Here, we present a neural model of decision making that can perform both evidence accumulation and action selection optimally. More specifically, we show that, given a Poisson-like distribution of spike counts, biological neural networks can accumulate evidence without loss of information through linear integration of neural activity and can select the most likely action through attractor dynamics. This holds for arbitrary correlations, any tuning curves, continuous and discrete variables, and sensory evidence whose reliability varies over time. Our model predicts that the neurons in the lateral intraparietal cortex involved in evidence accumulation encode, on every trial, a probability distribution which predicts the animal's performance. We present experimental evidence consistent with this prediction and discuss other predictions applicable to more general settings.

One of the most critical problems we face in the study of biological systems is building accurate statistical descriptions of them. This problem has been particularly challenging because biological systems typically contain large numbers of interacting elements, which precludes the use of standard brute force approaches. Recently, though, several groups have reported that there may be an alternate strategy. The reports show that reliable statistical models can be built without knowledge of all the interactions in a system; instead, pairwise interactions can suffice. These findings, however, are based on the analysis of small subsystems. Here, we ask whether the observations will generalize to systems of realistic size, that is, whether pairwise models will provide reliable descriptions of true biological systems. Our results show that, in most cases, they will not. The reason is that there is a crossover in the predictive power of pairwise models: If the size of the subsystem is below the crossover point, then the results have no predictive power for large systems. If the size is above the crossover point, then the results may have predictive power. This work thus provides a general framework for determining the extent to which pairwise models can be used to predict the behavior of large biological systems. Applied to neural data, the size of most systems studied so far is below the crossover point., Author Summary Biological systems are exceedingly complicated: They consist of a large number of elements, those elements interact in nonlinear and highly unpredictable ways, and collective interactions typically play a critical role. It would seem surprising, then, that one could build a quantitative description of biological systems based only on knowledge of how pairs of elements interact. Yet, that is what a number of studies have found. Those studies, however, focused on relatively small systems. Here, we ask the question: Do their conclusions extend to large systems? We show that the answer depends on the size of the system relative to a crossover point: Below the crossover point the results on the small system have no predictive power for large systems; above the crossover point the results on the small system may have predictive power. Moreover, the crossover point can be computed analytically. This work thus provides a general framework for determining the extent to which pairwise models can be used to predict the behavior of large biological systems. It also provides a useful heuristic for designing experiments: If one is interested in understanding truly large systems via pairwise interactions, then make sure that the system one studies is above the crossover point.

When a free electron laser (FEL) interaction occurs inside a waveguide, there are two possible operating frequencies at the two intersections of the beam line with the waveguide dispersion curve. Using a one-dimensional model valid for a low-gain oscillator, the mode competition between the high- and low-frequency waves is studied. It is found that the low-frequency wave can be nonlinearly suppressed, even when its linear growth rate is positive. When the frequency ratio is two, phase locking occurs and the nonlinear gain of the low-frequency wave can exceed its linear growth rate. >

The design and feasibility of a 1-MW continuous-wave (CW) free electron laser (FEL) oscillator are reviewed. The proposed configuration includes a short-period (I/sub w/ approximately 1 cm) planar wiggler, a sheet electron beam, a 0.5-1.0-MV thermionic electron gun, a hybrid waveguide/quasi-optical resonator, commercial DC power supplies, and a depressed collector. Cavity ohmic RF losses are estimated to be extremely low ( >

Recent psychophysical experiments indicate that humans perform near-optimal Bayesian inference in a wide variety of tasks, ranging from cue integration to decision making to motor control. This implies that neurons both represent probability distributions and combine those distributions according to a close approximation to Bayes' rule. At first sight, it would seem that the high variability in the responses of cortical neurons would make it difficult to implement such optimal statistical inference in cortical circuits. We argue that, in fact, this variability implies that populations of neurons automatically represent probability distributions over the stimulus, a type of code we call probabilistic population codes. Moreover, we demonstrate that the Poisson-like variability observed in cortex reduces a broad class of Bayesian inference to simple linear combinations of populations of neural activity. These results hold for arbitrary probability distributions over the stimulus, for tuning curves of arbitrary shape and for realistic neuronal variability.

It is widely acknowledged that an e+-e− linear collider in the TeV energy range will require linacs operating at frequencies higher than S-band. The reasons for this conclusion are reviewed. It is found, in particular, that the rf breakdown and wall heating limits are not the primary constraints for realistic designs; limits on ac-power consumption and/or on peak rf power per feed limit attainable accelerating gradients. Using a model for the physics at the interaction point and a power flow model for the linacs, we discuss some of the trade-offs involved in reaching a workable design. Two devices are considered: a 0.5 - 0.5-TeV collider and a 1.0 - 1.0-TeV collider, both operated at 10 GHz, both limited to an ac power consumption of 2x55 MW, and both designed for an interaction point luminosity of 1033 cm−2 sec−1. A tracking code, which includes wake-field effects, was used to find tolerable limits on injection jitter and on magnet misalignment in each case; these limits turn out to be very stringent; sub-micron magnet alignments are found to be required for minimal emittance growth, for example. A review of microwave sources capable of high power, high frequency operation is also given, and progress on the development of one such source, a gyroklystron, is reported.

A short-period-wiggler (period /spl equiv/ 1 cm), sheet-beam FEL has been proposed as a low-cost source of high-average-power (1 MW) millimeter-wave radiation for plasma heating and space-based radar applications. Recent calculations and experiments have confirmed the feasibility of this concept in such critical areas as rf wall heating, intercepted beam ("body") current, and high-voltage (0.5-1 MV) sheet-beam generation and propagation. Results of preliminary low-gain sheet-bearn FEL oscillator experiments using a field-emission diode and pulse line accelerator have verified that lasing occurs at the predicted FEL frequency. Measured start oscillation currents also appear consistent with theoretical estimates. Finally, we consider the possibilities of using a short-period, superconducting planar wiggler for improved beam confinement, as well as access to the high-gain, strong-pump Compton regime with its potential for highly efficient FEL operation.

At the University of Maryland high power microwave source development for use in linear accelerator applications continues with the design and testing of coaxial circuits for gyroklystron amplifiers. This presentation will include experimental results from a coaxial gyroklystron that was tested on the current microwave test bed, and designs for second harmonic coaxial circuits for use in the next generation of the gyroklystron program. The authors present test results for a second harmonic coaxial circuit. Similar to previous second harmonic experiments the input cavity resonated at 9.886 GHz and the output frequency was 19.772 GHz. The coaxial insert was positioned in the input cavity and drift region. The inner conductor consisted of a tungsten rod with copper and ceramic cylinders covering its length. Two tungsten rods that bridged the space between the inner and outer conductors supported the whole assembly. The tube produced over 20 MW of output power with 17% efficiency. Beam interception by the tungsten rods resulted in minor damage. Comparisons with previous non-coaxial circuits showed that the coaxial configuration increased the parameter space over which stable operation was possible. Future experiments will feature an upgraded modulator and beam formation system capable of producing 300 MW of beammore » power. The fundamental frequency of operation is 8.568 GHz. A second harmonic coaxial gyroklystron circuit was designed for use in the new system. A scattering matrix code predicts a resonant frequency of 17.136 GHz and Q of 260 for the cavity with 95% of the outgoing microwaves in the desired TE032 mode. Efficiency studies of this second harmonic output cavity show 20% expected efficiency. Shorter second harmonic output cavity designs are also being investigated with expected efficiencies near 34%.« less

It has been known for >30 years that neuronal spike trains exhibit correlations, that is, the occurrence of a spike at one time is not independent of the occurrence of spikes at other times, both within spike trains from single neurons and across spike trains from multiple neurons. The presence of these correlations has led to the proposal that they might form a key element of the neural code. Specifically, they might act as an extra channel for information, carrying messages about events in the outside world that are not carried by other aspects of the spike trains, such as firing rate. Currently, there is no general consensus about whether this proposal applies to real spike trains in the nervous system. This is largely because it has been hard to separate information carried in correlations from that not carried in correlations. Here we propose a framework for performing this separation. Specifically, we derive an information-theoretic cost function that measures how much harder it is to decode neuronal responses when correlations are ignored than when they are taken into account. This cost function can be readily applied to real neuronal data.

At the University of Maryland, we are exploring the suitability of gyroklystrons for linear collider applications. In this paper we discuss the design of a two-cavity, coaxial gyroklystron that is designed to produce over 100 MW in 1 /spl mu/s pulses at the second harmonic of the cyclotron frequency. We detail modifications that will be required of our test facility, present the simulated performance of a 400 MW, 500 kV single-anode magnetron injection gun, and describe the design of the coaxial microwave circuit. >

We report the operating characteristics of a sequence of two-cavity second harmonic K-band gyroklystrons. The TE/sub 011/ input cavity is driven near 9.88 GHz. The first four tubes utilized smooth transition output cavities which resonate at twice the drive frequency in the TE/sub 021/ mode. Peak powers in excess of 30 MW were achieved with efficiencies greater than 28% and large signal gains of 27 dB. The results of an abrupt transition complex TE/sub 01/, TE/sub 02/ mixed mode output cavity are also discussed, as well as the prospect of placing a coaxial insert into the highest power tube. >

Summary form only given. The Naval Research Laboratory is currently investigating gyro-twystrons as high-power, wideband sources for millimeter wave radars. For many radar applications, the gyro-twystron configuration represents a suitable compromise between the high power, narrow band gyro-klystron and the lower power, broadband gyro-TWT. A non-linear simulation code was used to design two gyro-twystrons, the first in Ka-band and the second in W-band. For the Ka-band device, the interaction between the TE/sub 01/ mode and a 65 kV, 7 A electron beam with /spl alpha/=1.3 was studied. The rms perpendicular electron velocity spread was assumed to be 4.5%. The simulation results indicate that an output power of 123 kW, corresponding to an efficiency of 27%, can be achieved. The 3 dB instantaneous bandwidth, which extends from 34.6 GHz to 36.2 GHz, is 4.5%. Next, the non-linear code was used to design a W-Band gyro-twystron. The interaction mode was, once again, the TE/sub 01/. A 65 kV, 4 A beam with /spl alpha/=1.3 and an RMS perpendicular velocity spread of 4.5% was used. The simulations show that 34 kW output power, corresponding to 13% efficiency, can be realized with a 3 dB instantaneous bandwidth of 4.3% (91 GHz to 95 GHz). The theoretical bandwidths were obtained assuming a frequency independent bunching parameter.

Summary form only given, as follows. Experiments on a two-cavity gyroklystron are underway to demonstrate a 140 kW, 35 GHz gyroklystron amplifier, operating at a fundamental beam cyclotron mode and a TE/sub 011/ cylindrical cavity mode. A high power electron beam of 70 kV, 6.6 A is produced from a magnetron-injection-gun which is optimally designed for the TE/sub 01/ mode at 35 GHz. Drift tubes consisting of lossy ceramic rings (80% BeO, 20% SiC) are designed to suppress undesired oscillations. A drive power is injected into the first cavity through a multi-hole coaxial coupler. A capacitive probe is placed directly before the input cavity to measure the beam velocity ratio. Large signal non-linear calculations predict a peak efficiency of 30% (extracted power=140 kW) and a saturated gain of 20 dB over a 0.3% bandwidth at /spl alpha/=1.5, /spl Delta/v/sub Z//v/sub Z/=20% at 13.3 kG and Q/sub 1/=Q/sub 2/=200. Design parameters and initial hot-test results of the amplifier will be presented.

We present a summary of the results achieved with our 30 MW gyro-amplifier test facility during the past several years. Twenty-one different tube configurations have been tested in 5 major categories: fundamental, second harmonic, and third harmonic gyroklystrons, as well as fundamental and second harmonic gyrotwystrons. We also present designs of experiments with fundamental and second harmonic coaxial gyroklystrons which are expected to produce output powers near 100 MW at 8.568 and 17.136 GHz, respectively, in the near future. >

The authors are working to develop high-power gyroklystron amplifiers with properties suitable for driving future electron-positron colliders. Initial experiments have involved a sequence of six different two-cavity configurations, operating in the TE/sub 01/ mode near 9.85 GHz. Each design was modified in accordance with its predecessor's performance to either improve tube stability or enhance amplifier operation (or both). The energy is derived from a rotating beam in 300-440 kV, 100-200 A, 1- mu s pulses. With perpendicular-to-parallel velocity ratios near one, the studies have culminated in peak powers approaching 24 MW, with efficiencies and large-signal gains of 33% and 34 dB, respectively. The experimental setup is described, and amplifier performance of the most recent configuration is discussed. >

The theoretical design of a 20-GHz second-harmonic gyroklystron amplifier at a voltage near 500 kV is presented. The design of the output cavity and the nonlinear gain and efficiency are discussed. It is shown that second-harmonic operation with bunching at the fundamental frequency is feasible. With a velocity spread of 10%, numerical simulations show an output power of 15 MW (corresponding to an efficiency of 22%) at 20 GHz. The gain was about 16 dB. Output cavities that can be used to prevent feedback of the output cavity signal into the input cavity and to guarantee that the output will be in a pure mode are presented. >

The authors present experimental results of a 10 GHz TE/sub 01/ mode three-cavity gyroklystron. The beam is produced by a pulse line modulator and magnetron injection gun, which can operate to 433 kV and 225 A with 1 mu s flat-top and at a repetition rate of 3 Hz. Microwave power was measured by a mode-selective directional coupler and a flowing methanol calorimeter. Mode purity was determined by a large anechoic chamber. Initial testing of the first three-cavity circuit has produced a peak power of 23 MW with an efficiency of 27% and pulse energy of 36 J. A maximum gain of 39 dB at a peak power of 21 MW was obtained. >

Summary form only given. A TE/sub 01/ gyroklystron experiment in the 9.7-10.0 GHz range is discussed. The target microwave power of 30 MW is to be derived from the interaction of a 500 kV, 160-A beam with a two-cavity circuit. The first tube was tested in spring 1990 and was plagued by a multitude of spurious oscillations. The most damaging oscillations included 'whole tube' modes and modes in the beam tunnel between the electron gun and the input cavity. Limited amplification studies were performed at 175 kV and produced only tens of kilowatts at nearly zero gain. The second tube featured additional beam tunnel loss and a new drift tube configuration, and was tested in summer 1990. Enhanced stability allowed amplification studies up to 350 kV, which resulted in output powers in excess of 1.5 MW with gains of 16 dB and efficiencies of approximately 5%. A spurious mode near 9.85 GHz existing in the output waveguide adjacent to the final cavity severely limited the available magnetic field range. A third tube, with a redesigned output waveguide, will undergo testing by September 1, 1990. >

A four cavity W-band gyroklystron amplifier experiment is currently underway at the Naval Research Laboratory. The gyroklystron has produced 67 kW peak output power and 28% efficiency in the TE/sub 011/ mode with a 55 kV, 4.3 A annular electron beam. The full width half maximum (FWHM) bandwidth is greater than 460 MHz. Small signal and saturated gains of 36 dB and 29 dB, respectively, have been observed. The amplifier is zero drive stable and the limiting oscillation is the TE/sub 011/ operating mode in the output cavity. Experimental results are in good agreement with theoretical predictions.