Multi-bit Flip Optimization for Ising Model Search Speed
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional optimization apparatuses using Ising energy functions for solving multivariable optimization problems face a slowdown in searching for an optimal solution due to the low probability of bit flipping at lower temperatures, especially when implementing simulated annealing with stochastic search methods.
Innovation Solution
An optimization apparatus that calculates changes in energy for multiple bit flips simultaneously, using first and second local field values to select bits for update based on thermal excitation energy and random numbers, allowing exploration of neighborhood states beyond single-bit flips, thereby increasing the speed of finding optimal solutions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If conventional optimization apparatuses implement simulated annealing by accepting only single bit flips at a time, then the stochastic search process maintains simplicity, but the speed of searching for optimal solutions decreases when temperature is lowered
Solution Approach 1:
The patent extends the search from single-bit flips to multi-bit flips by introducing a parameter n (where n ≥ 2) that represents the number of bits to flip simultaneously. This dimensional change allows the system to explore neighborhood states with Hamming distance n, transforming the search process from one-dimensional (single bit) to multi-dimensional (multiple bits), thereby resolving the contradiction between operational simplicity and search speed.
Solution Approach 2:
The patent segments the bit flip operation into independent units where each unit can be flipped individually or in combination with other bits. By calculating energy changes for multiple bit flips in parallel and selecting from these segmented options, the system maintains the simplicity of individual bit operations while achieving the productivity benefits of multi-bit exploration.
2Reliability
If the probability of obtaining a bit to be flipped is low at lower temperatures, then the simulated annealing process maintains proper thermal behavior, but the search speed slows down significantly
Solution Approach 1:
By changing from single-bit to multi-bit flips, the patent increases the effective search space exploration per iteration. Even though the probability of accepting any individual bit flip remains low at low temperatures, the system compensates by considering combinations of multiple bits, thereby reducing the time lost to slow search progression.
Solution Approach 2:
The patent calculates energy changes for multiple potential bit flips in advance before performing the selection. By pre-computing the energy differences for multiple bits and their combinations, the system prepares potential moves ahead of time, reducing the time needed during the actual search process when temperature is low.
3Productivity
If multi-bit flips are considered simultaneously, then the search speed and accuracy improve, but the device complexity increases
Solution Approach 1:
The patent segments the complex multi-bit energy calculation into multiple simpler single-bit energy difference calculations. By calculating the energy change for each bit independently and then combining these results to evaluate multi-bit flips, the system reduces the complexity of individual calculation circuits while achieving the productivity benefits of multi-bit search.
Solution Approach 2:
The patent introduces a new dimension to the energy calculation by considering combinations of multiple bits simultaneously. Rather than calculating all possible multi-bit energy changes from scratch, the system builds upon single-bit energy differences, adding a combinatorial dimension that improves search efficiency without proportionally increasing circuit complexity.
Data Source
AI summary
In an optimization apparatus, calculation circuits individually calculate a change in energy based on first local field values each associated with one of n bits (n is an integer greater than or equal to 2) amongst a plurality of bits corresponding to spins in an Ising model, values of the n bits, and one or more weight values representing strength of interaction among the n bits. The change in energy is caused by flips of the n bits. An update bit selection circuit selects the n bits for which value update is accepted, based on the magnitude relationship between thermal excitation energy and each of the calculated changes in energy. An update circuit flips the n bits, and also updates, based on the flips of the n bits, second local field values including the first local field values and each associated with the plurality of individual bits.