Abstract: A computational circuit for a multi-value addition comprising a parallel adder, an output adder, a quantizing portion and a logic conversion portion. Addition circuits in the above adders and thresholding circuits in the above quantizing portion consist of voltage-driven circuits including capacitive couplings.

Abstract: A random number generator for generating random numbers. The generator may include two similarly implemented random signal sources, with each source generating a random signal. A difference signal of the two random signals is generated with the result that any common interference signal components are eliminated.

Abstract: The reliability of solutions of simultaneous algebraic equations with integral coefficients is improved and high-speed information processing is realized. A method processes information wherein, in a model in which restrictions among parameters are given by simultaneous algebraic equations with integral coefficients, when the simultaneous equations have only a finite number of solutions, zero points of a polynomial set F with integral coefficients, which represents the simultaneous equations and is described in a memory, are represented by rational expressions based on zero points of a one-variable polynomial regarding one variable so that the information is represented using the representation of rational expression. The method processes information by choosing a term order, calculates a Grobner basis, calculates a minimum polynomial f1, and controls a digital processor to find solutions represented by rational expressions.

Abstract: A desired programmable complex analog bandpass filter is replaced by a "black box" that receives an analog continuous-time signal and delivers an analog continuous-time signal, just as the analog filter would, but internally realizes the programmability with a programmable discrete-time system. The transfer function of the "black box" is the same as the transfer function of the analog filter. The "black box" includes an anti-aliasing filter, an A/D-converter, a complex digital bandpass filter (16), a D/A-converter (18) and an anti-imaging filter (20).

Abstract: The cut-off frequency and boost value of an analog filter on a read channel is set to initial values When the data is read from a disk, the read data is subjected to PR equalization at an FIR filter. At that time, a learning function provided on the FIR filter adjusts the tap coefficient so that an equalization error may be minimized. The cut-off frequency and boost value of the analog filter are modified until the adjusted tap coefficient has gone in a permitted range. After the tap coefficient of the FIR filter has gone in the permitted range, the cut-off frequency and boost value are stored in an EEPROM so that they can be used in a normal mode.

Abstract: A finite impulse response filter uses rotating tap weights that are connected in turn to a succession of sample/hold (S/H) cells in each of which is held a separate successive sample of an analog balanced signal being filtered for an interval long enough to tap to each tap weight in one rotation. Each of the S/H cells is a switched-current memory cell and includes a primary negative feedback transconductance differential amplifier and a primary pair of storage capacitors and a secondary positive feedback transconductance differential amplifier and a secondary pair of storage capacitors. Charge injection errors created in each primary amplifier are replicated in an associated secondary amplifier and used to cancel these errors to provide higher fidelity samples for summing for forming the output of the filter.The tap weights are provided by binary-ratioed resistors provided by transistors of binary-ratioed transconductances.