This paper proposes a lossless coding method for the compression of computer files of floating-point or fixed-precision numbers. The method is adaptive in the sense that it requires no prior knowledge about input data. Although it is quite simple and all that needed is the incremental parsing technique by Ziv and Lempel, the proposed method compresses well any i.i.d. sequence of numerical data generated from a source with a smooth distribution. In order to evaluate the performance and the convergence property, a bitwise equivalent model is introduced, which combines the Ziv-Lemple-type data compression methods with the probabilistic framework. The model shows that, for any sufficiently long i.i.d. sequence, the proposed method attains the entropy with only a loss of 0.36 bits per word or so. Computer-simulation results are also presented in support of this evaluation.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Hidetoshi YOKOO, "A Lossless Coding Algorithm for the Compression of Numerical Data" in IEICE TRANSACTIONS on transactions,
vol. E73-E, no. 5, pp. 638-643, May 1990, doi: .
Abstract: This paper proposes a lossless coding method for the compression of computer files of floating-point or fixed-precision numbers. The method is adaptive in the sense that it requires no prior knowledge about input data. Although it is quite simple and all that needed is the incremental parsing technique by Ziv and Lempel, the proposed method compresses well any i.i.d. sequence of numerical data generated from a source with a smooth distribution. In order to evaluate the performance and the convergence property, a bitwise equivalent model is introduced, which combines the Ziv-Lemple-type data compression methods with the probabilistic framework. The model shows that, for any sufficiently long i.i.d. sequence, the proposed method attains the entropy with only a loss of 0.36 bits per word or so. Computer-simulation results are also presented in support of this evaluation.
URL: https://global.ieice.org/en_transactions/transactions/10.1587/e73-e_5_638/_p
Copy
@ARTICLE{e73-e_5_638,
author={Hidetoshi YOKOO, },
journal={IEICE TRANSACTIONS on transactions},
title={A Lossless Coding Algorithm for the Compression of Numerical Data},
year={1990},
volume={E73-E},
number={5},
pages={638-643},
abstract={This paper proposes a lossless coding method for the compression of computer files of floating-point or fixed-precision numbers. The method is adaptive in the sense that it requires no prior knowledge about input data. Although it is quite simple and all that needed is the incremental parsing technique by Ziv and Lempel, the proposed method compresses well any i.i.d. sequence of numerical data generated from a source with a smooth distribution. In order to evaluate the performance and the convergence property, a bitwise equivalent model is introduced, which combines the Ziv-Lemple-type data compression methods with the probabilistic framework. The model shows that, for any sufficiently long i.i.d. sequence, the proposed method attains the entropy with only a loss of 0.36 bits per word or so. Computer-simulation results are also presented in support of this evaluation.},
keywords={},
doi={},
ISSN={},
month={May},}
Copy
TY - JOUR
TI - A Lossless Coding Algorithm for the Compression of Numerical Data
T2 - IEICE TRANSACTIONS on transactions
SP - 638
EP - 643
AU - Hidetoshi YOKOO
PY - 1990
DO -
JO - IEICE TRANSACTIONS on transactions
SN -
VL - E73-E
IS - 5
JA - IEICE TRANSACTIONS on transactions
Y1 - May 1990
AB - This paper proposes a lossless coding method for the compression of computer files of floating-point or fixed-precision numbers. The method is adaptive in the sense that it requires no prior knowledge about input data. Although it is quite simple and all that needed is the incremental parsing technique by Ziv and Lempel, the proposed method compresses well any i.i.d. sequence of numerical data generated from a source with a smooth distribution. In order to evaluate the performance and the convergence property, a bitwise equivalent model is introduced, which combines the Ziv-Lemple-type data compression methods with the probabilistic framework. The model shows that, for any sufficiently long i.i.d. sequence, the proposed method attains the entropy with only a loss of 0.36 bits per word or so. Computer-simulation results are also presented in support of this evaluation.
ER -