Stochastic Random Number Generation via Weighted Coinflip Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current pseudo-random number generators (PRNGs) are inadequate for applications requiring high-quality random numbers, especially in parallel architectures, as they produce random numbers from a uniform distribution, which often necessitate additional computation to convert to the required distribution, and struggle with the serial operation in highly parallel systems.
Innovation Solution
The method involves using weighted coinflip devices to directly generate random numbers from a target distribution by performing coin flips, where the weights are determined by a function, and converting these coin flips into a binary representation of random numbers, leveraging stochastic devices like magnetic tunnel junctions and tunnel diodes integrated into a neuromorphic architecture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If pseudo-random number generators (PRNGs) are used to generate random numbers, then ease of generation and utility in verification is improved, but additional computation is required to convert from uniform distribution to target distribution, and serial operation creates complexities in parallel architectures
Solution Approach 1:
The patent replaces the mechanical/deterministic PRNG system with a stochastic hardware system using coinflip devices. These devices physically generate random numbers directly from target distributions through stochastic processes, eliminating the need for deterministic algorithms and post-generation conversion computations.
Solution Approach 2:
The patent changes the fundamental parameter of random number generation from deterministic sequential generation to stochastic parallel generation. By using coinflip devices with adjustable weights corresponding to target distribution probabilities, the system generates numbers directly in the desired distribution without conversion.
2Extent of automation
If pseudo-random number generators (PRNGs) are used, then repeatability through seeding is improved, but quality of random numbers for stringent applications like cryptography deteriorates
Solution Approach 1:
The patent introduces dynamic control over the coinflip devices through weight adjustment mechanisms. The weights can be configured to match target distribution probabilities while maintaining the ability to reproduce specific distributions deterministically when needed, thus achieving both quality and repeatability.
3Device complexity
If pseudo-random number generators (PRNGs) operate in serial mode, then simplicity of implementation is improved, but productivity in parallel architectures deteriorates
Solution Approach 1:
The patent segments the random number generation process into multiple independent coinflip devices that can operate in parallel. Each device handles a portion of the generation task, allowing simultaneous production of multiple random numbers and dramatically increasing throughput in parallel architectures.
Solution Approach 2:
The patent creates a universal stochastic computing platform where coinflip devices can generate random numbers from any target distribution by adjusting weights. This multi-functional approach replaces multiple specialized PRNG implementations with a single parallelizable stochastic system.
Data Source
AI summary
A method for probabilistic computing is provided. The method comprises specifying a target distribution for a computational model, wherein the target distribution is defined by a function. A number of coin flips are performed with a number of weighted coinflip devices, wherein weights for the coinflip devices are determined by the function. The number of coinflips are then converted to a random number from the target distribution according to outputs of the weighted coinflip devices, wherein a circuit uses the coin flips as inputs to randomly activate bits in a binary representation of the random number.