Floating-Point Compression Using Integer Prediction for CAD Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for compressing CAD files with floating-point data fail to ensure stability across different platforms, leading to inconsistent results due to floating-point arithmetic differences, and often result in inefficient compression rates due to the use of quantization and prediction schemes that limit the accuracy of predictions.
Innovation Solution
A computer-implemented method that computes a floating-point predicted value by performing operations on integers corresponding to the sign, exponent, and mantissa of actual values, storing a bit sequence representative of the difference between the target actual value and the predicted value, and iteratively applying these steps to efficiently compress and decompress numerical data, ensuring lossless and stable compression across platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If quantization and prediction schemes are used to compress floating-point data, then compression rate is improved, but manufacturing precision and measurement precision deteriorate due to loss of accuracy
Solution Approach 1:
The patent changes the parameter representation from direct floating-point storage to differential encoding where only the difference between consecutive values is stored. This transformation allows for higher compression rates while maintaining full precision of the original data, resolving the contradiction between compression efficiency and data accuracy.
Solution Approach 2:
The patent segments the floating-point data into individual elements that can be independently compressed using differential encoding. By processing each floating-point value separately and storing only the difference from the previous value, the system achieves efficient compression without sacrificing precision, as each segment maintains its full accuracy.
2Ease of manufacture
If floating-point arithmetic operations are performed during compression, then compression functionality is achieved, but stability across different platforms deteriorates due to floating-point differences
Solution Approach 1:
The patent substitutes floating-point arithmetic operations with integer-based differential encoding. Instead of performing floating-point calculations during compression, the system stores only the difference between consecutive floating-point values as an integer offset. This replacement eliminates platform-specific floating-point arithmetic issues while maintaining compression functionality, as integer operations are deterministic across all platforms.
Solution Approach 2:
The patent introduces an intermediary differential value that mediates between the original floating-point data and the compressed representation. By storing only the difference (delta) between consecutive values rather than the full floating-point values, the system achieves compression while avoiding unstable floating-point operations during the compression process itself.
3Quantity of substance
If differential encoding is used to compress floating-point values, then storage size is reduced, but device complexity increases due to additional processing requirements
Solution Approach 1:
The patent changes the storage parameter from full floating-point values to differential offsets. Instead of storing complete floating-point numbers, the system stores only the difference from the previous value, significantly reducing storage requirements. The processing complexity is minimized by using simple subtraction operations to generate the differential values.
Data Source
AI summary
The invention relates to a computer-implemented method for compressing numerical data comprising a structured set of floating point actual values. A floating point value is defined by a sign, an exponent and a mantissa. The method comprises computing a floating point predicted value related to a target actual value of the set. The computing includes performing operations on integers corresponding to the sign, to the exponent and/or to the mantissa of actual values of a subset of the set. The method also comprises storing a bit sequence representative of a difference between integers derived from the target actual value and the predicted value. Such a method is particularly efficient for reducing the storage size of a CAD file.