Liquid crystal light valves (LCLVs) have found application in many proposed and demonstrated optical computing systems. In many cases the polarization state of the illumination is used to encode information, e.g., symbolic substitution and neural networks. In these applications it is often desired that two linear and orthogonal states of polarization result from the devices used. This is the form of polarization coding used in this paper. Any deviations from these desired polarization states will results in signal crosstalk within the system. Typically, the light reflected from the light valve is elliptically polarized. By adjusting the drive conditions or setting the light valve at a fixed azimuth, the degree of ellipticity can be minimized. However, the rotation of the major axis is usually less than 90 degree(s). The authors characterize three different types of LCLVs based on twisted nematic, aligned nematic, and ferroelectric (smectic C) liquid crystals. The optical activity and birefringence of the devices as a function of write light intensity are being studied. System crosstalk resulting from any deviations from the desired characteristics is discussed.

We analyze the influence of input information coding in an optical Van Der Lugt correlator for pattern recognition. When an image is generated by a spatial light modulator (SLM) on a coherent beam, its phase and amplitude are strongly coupled. This is particularly the case with the available liquid crystal SLM studied in this paper. After illustrating typical experimental results obtained by coding with such a SLM (thin film transistor twisted nematic), three coding techniques are quantitatively analyzed: phase and amplitude, phase only, and amplitude only coding. Their influence on the correlation peak is studied in terms of optical efficiency, sharpness, and, probably more important, noise robustness of the correlation peak.

Theoretical simulation shows that it is possible to optimize computer-generated holograms performing simple interconnect functions for limited coherence illumination. Further simulation indicates that a development of a silicon backplane spatial light modulator (SLM) can be used as the hologram recording device. A model of a programmable beamsplitter network is successfully modelled, and results are presented.

The construction of digital optical processors based on the cellular logic image processor (CLIP) architecture is discussed. Both a single-channel processor and a parallel version incorporating 256 information channels have been constructed. The single channel version of the processor allows eight different combinatorial logic processes to be carried out under electronic control and can be programmed in real time. Several algorithms including pattern recognition, byte comparison, full addition and subtraction have been implemented with this machine. The 256 channel version operates similarly to the single channel version except that a reduced instruction set internal processor with four selectable logic processes is used. A nearest neighbor interconnect provides the communication required between the different information channels. More advanced processing capability can be achieved with the introduction of such non-local interconnects as shuffle networks. Results and simulations obtained with these processors are presented. Advances in the various components of the O- CLIP circuit, future goals, and potential application are also discussed.

The digital optical computing program within the National Science Foundation Engineering Research Center for Optoelectronic Computing Systems has as its specific goal research on optical computing architectures suitable for use at the highest possible speeds. The program can be targeted toward exploiting the time domain because other programs in the Center are pursuing research on parallel optical systems, exploiting optical interconnection and optical devices and materials. Using a general purpose computing architecture as the focus, the authors are developing design techniques, tools, and architectures for operation at the speed of light limit. Experimental work is being done with the somewhat low speed components currently available but with architectures which will scale up in speed as faster devices are developed. The design algorithms and tools developed for a general purpose, stored program computer are being applied to other systems such as optically controlled optical communications networks.

The optical module described is designed to perform cascadable optical parallel logic operations, using 2-D binary registers consisting of two optically addressed ferroelectric liquid crystal spatial light modulators. The high contrast ratio and bistability of the device are particularly useful in the construction of optical logic units. This paper illustrates the cascading abilities of such a device in performing a parallel optical XOR logic operation.

A motionless head 2-D parallel readout system for optical disks is presented. The system is designed to read data blocks encoded as 1-D Fourier holograms distributed radially on the disk active surface. Such systems offer several advantages: high data rates, low retrieval times, and simple implementation. It is used as the secondary storage of a high performance optoelectronic associative memory system.

This paper describes recent progress in the development of a new class of spatial light modulator (SLM). These new SLMs modulate light by the interaction of some active material with a high intensity evanescent field generated by surface plasmon resonance. Such devices have the potential for substantial advantages over conventional SLMs, including higher speed and better response uniformity, as well as high sensitivity in devices with thin active layers. A new optically addressed plasmon device, based on a thin amorphous silicon/liquid crystal sandwich structure, has been developed and tested. The performance characteristics compare favorably with those of conventional liquid crystal SLMs in terms of resolution and speed. The design of more advanced devices based on higher performance ferro-electric and electroclinic liquid crystals is now in progress; in particular, the special pseudo-plasmon modes found in highly birefringent materials, and the application of these to modulation, have been analyzed. Surface plasmon SLMs using electro-optic effects in semiconductor active layers are also discussed.

Theoretical studies are given for a nonlinear Fabry-Perot resonator filled with a birefringent medium or a gyrotropic medium. The results mapped onto Poincare's spheres show the changes of the output polarization states due to the changing input intensities.

Bistabilities of photocurrent and photoluminescence at the low-temperature impurity breakdown suggests all elements necessary for optoelectronic and optical processing. It was found that such elements may possess record low energy and power characteristics with a good speed of operation. However, the necessity of using a low temperature (for many semiconductors it is a helium temperature) seriously limits the range of applications of these elements.

The pattern recognizing power of a double-layer hetero-associative neural network is discussed in some detail and demonstrated using a set of 256 Chinese characters. The results show that a pattern, unrecognizable to the human eye, can be recalled as being a corrupted version of a particular character from this set.

A compact optical coprocessor dedicated to the edge extraction on binary images is presented. Four cellular automata using optical interconnections implementing an OR logical function and optical photothyristors acting as inverting and noninverting optical gates are proposed. This machine could reasonably process 1024 X 1024 images with present technologies.

Optical computer designs must implicitly or explicitly allow for power budgeting to compensate for crosstalk and loss. This paper develops algorithms for calculating the system crosstalk and power loss in optical systems, using a graph theoretic system model. These algorithms form the basis for an optical systems design methodology in which the designer can develop a system design that employs lumped delays, and that assumes zero device delay, crosstalk and loss, and then employ the algorithms developed in this paper to estimate actual system delay distribution, crosstalk, and power loss.

All-optical switching at power levels down to 4(mu) W using the large thermal dispersive nonlinearity present in nematic liquid crystals is presented. Parallel operation is demonstrated in the form of a 15 X 15 array memory, and design criteria for three-port logic operation are considered with regard to circuitry requirements such as gain, speed, and fan-out. The implementation of an all-optical flip-flop consisting of coupled three-port logic gates is also described.

The photon echo application for optical digital processing of discrete images by the method of managing operators was examined. Photon echo is the pulse of light which is emitted spontaneously from a system of atoms previously irradiated by two or more coherent light pulses with length and power of 102divided by10-1 ps and 104divided by105 kW, respectively. These systems of resonant atoms are crystals with various paramagnetic activators. By using photon echo, the resonant systems are able to store and retrieve wave-front and temporal form of one of the excitation pulses. The photon echo application for information processing allows one to solve a number of problems with high efficiency. Different photon echo generating conditions can be utilized practically in all processor components, which creates certain prospects for using this effect in optical computing purposes.

High bandwidth computing makes equally high demands upon the speed of data processing and the speed of data flow. Consequently, in high bandwidth computing systems at several levels, fast and highly parallel interconnects are necessary for both clock and data distribution in order to avoid transfer-bound computing. For the present time, data processing has still to be based upon the use of electronic components. However, realization of fast and highly parallel interconnects is no longer possible without application of optical (and electrical) interconnects in combination with electro-optic and opto-electronic elements. Realization problems are quite different at system, inter-chip and intra-chip level. High bandwidth computing requires at system level application of powerful, reconfigurable interconnects. Consequently, the feasibility of optical interconnects at the system level is closely related to the system interconnect topology. The advantages of the use of optical interconnects for the purpose of data distribution has given rise to consideration of the feasibility of replacing electrical power distribution by means of electrical links with optical power distribution through optical links.

Current limitations for interconnection and control in electronic integrated circuits, from the architectural and technological point of view, are presented. Different kinds of interconnection problems within a chip are briefly presented, according to a specific classification. Then, the authors show how they have incorporated optoelectronic devices using a standard CMOS technology to resolve one interconnection problem: the chip clock distribution and control. Some advantages and limitations of this technique used to sequence and control a 8*8 retina are presented. The authors discuss how these experiments, with other techniques used to output data, can provide a possible approach to an optical bus.

For fast computer-internal data transmission between different components, two new optoelectronic parallel bus systems have been developed. They are particularly appropriate for use as backplanes. The first optical backplane structure consists of a lot of circular transparent plates with a refractive index n1 and covered on both sides by a material with a refractive index of n2 (n21). The plates are optically isolated and arranged in a stack. This bus system is referred to as the optical parallel plate stack (OPPS). The optical bus signals are coupled in and out of the stack at the periphery edges of each individual plate. The second optical backplane structure consists of strips of a light-transmitting medium with, for instance, a rectangular cross section. The optical transmitting medium (refractive index n1) is cladded by a coating (refractive index n2 and n21). The strips separated by an optical isolation are put together, thus building a plate. This system is referred to as an optical parallel strip plate (OPSP). The optical signals are coupled in and out of the strips through the narrow surfaces of the strips. Study results concerning the radiated power and pulse dispersion of received optical signals show that data-transfer rates of more than 1.5 Gbit/s per plate channel and strip channel, respectively, can be realized at a bus length of 1 m.

There is real opportunity for optical computing, especially free-space optical communication, to improve the performance of supercomputers that exploit parallel processing techniques. In order to seize this opportunity, this paper reviews some of the needs of parallel architectures that appear hard to satisfy with standard electronic techniques and that may be more amenable to optics. The ability to vary the focus of a free-space optical beam lets one choose between sending a signal between two points in a narrow focus or to broadcast a signal from one source point to many target points with a wide focus. Electronic communication has mainly considered the former, but it is known that a little bit of broadcasting can go a long way in improving the performance of many applications. Even when one considers the power consumption (e.g., broadcast is done at a lower data rate) performance may still be improved. The standard techniques for broadcasting in parallel processors and how they are employed in current applications are described. Using these results, it becomes more clear where optics may help.

The modern tendenci es i n the techni ques of acoustoopti c ( AO) data proceasing stinulate the technology development of the rnultichannel AO devices eleroental base, in the first turn nultichannel Bragg modulators' ' Z The demands made to them in the composition of the high-production signal and algebraic processors can be deVided into two groups. The demands of the large working band and the modulators high diffraction efficienty can be attributed to the first one. The realization of high indices on these parameters is based in the main on the technique achievements of the one channel AO modulators , progress i n worki ng out of whi ch i s defi ned by the successes in synthesizing of the new high effective AO materials, or achivernents in looking for the high effective cut-offs of the known crystals. The given criterion can be indicated by the value of AO quality coefficient M of the material for isotropic or anisotropic wide-band diffraction. Such important parameters as the admissible level of the interchannel coupling, the identity degree of the amplitude and phase-frequency channel characteri sti cs , quantity and densi ty of channel arrangement on the crystal can be attributed to the second group. The complex of these requl rements causes the more hard cri ten a i n selecting the material of the modulator sound wire and the AOl geometry, adding them by the physical parameters, playing the secondary role for one-channel modulators. To those in particular there have been attn buted the coeffi ci ents of the quadrati c acousti c ani sotropy showing the relative increasing or decreasing of the beam ( energetic) sound di vergence on compari son wi th the di ffracti on one. The value of these coefficients can greatly influence on the density of channel arangeroent in some cases. For example in widely used in acoustooptic mnodulators on TeO for slow shi fted waves near di recti on (11 0) the acoustic beam divergence forces in 50 times on comparison with the diffraction one. This circumstance limits essentially the use of the gi yen cut-off s TeO for building of the multichannel AO processor. In the given work we examine the main physical and technological aspects in the domain of development of wide-band multichannel microwave AO modulators of two different types: 1 - modulators on the crys— tal LiNJ on the base anisotropic AO interaction in YZ plane and the excitation of the acoustic waves on the crystal surface by the slotted transducer; 2 - isotropic modulators on the crystal TeO i-cut-ting with piezotransducer on the crystal LiNbO plates. The technology of the modulators with the surface excitation of the acoustic waves is more simple than the technology of the modulators with the plate or film transducer, but demands optimization of the interaction geometry with the joint calculation acoustooptical and piezoelectrical properties. the base of expression taking into account optical , acoustic and piezoelectrical anizotropy of the crystals by the numerical methods there have been analyzed the angular dependences of the AO quality coefficient M for isotropic and anisotropic geometries AO interaction in the crystal LiNW . There have been determined that the absolute maximum of the value M is achieved for the wide-band anisotropic light diffraction on the slow shifted waves of the IZ crystal plane. In the maximum M region the slow shifted waves possess the cross piezoactivity, moreover the coefficient of the electromechanical coupling L. for the given type of piezoactivity also reaches the absolute maxinum, that gives the possibility sufficiently effectively excite these waves on the crystal plane and develop simple from the technological point of view one channel and multi channel anisotropic microwave AO modulators. The analysis of the acoustooptical and acoustic anisotropy of the crystal TeO allows us to suppose that the most acceptable geometry of the AC interaction for building rnultichannel Bragg modulators correspond to the isotropic light diffraction on the longitudinal waves i n the di recti on ( 001 ) . This geometry is characteri zed by the peak anisotropy M relatively the acoustic wave deviation from the direction (001) in the sector 100 and by the small values of the quadratic acoustic anisotropy coefficients W.

An optoelectronic hybrid optical array logic system is experimentally constructed as an instance of the optical parallel array logic system (OPALS). In the experimental system, a coherent 2-D correlator with holographic filters and image encoder/decoder based on fully parallel electronic circuits are applied. Holographic filters for several kinds of simple processing are optically fabricated and correct operation of the system is verified. In the experimental system, 3 by 3 pixels are processed in parallel at the rate of about 278 operations per second. Sixteen logical functions and iterative operations are demonsirated.

This paper suggests a new class of digital logic. OptiComp has focused on a digital optical logic family in order to capitalize on the inherent benefits of optical computing, which include (1) high FAN-IN and FAN-OUT, (2) low power consumption, (3) high noise margin, (4) high algorithmic efficiency using 'smart' interconnects, (5) free space leverage of GIBP (gate interconnect bandwidth product). Other well-known secondary advantages of optical logic include (but are not limited to) zero capacitive loading of signals at a detector, zero cross-talk between signals, zero signal dispersion, minimal clock skew (a few picoseconds or less in an imaging system). The primary focus of this paper is to demonstrate how each of the five advantages can be used to leverage other logic family performance such as GaAs; the secondary attributes will be discussed only in the context of introducing the DOC III architecture.