Langevin granulometry of the particle size distribution

Attila Kákay, M. W. Gutowski, L. Takacs, V. Franco, L. Varga

Research output: Contribution to journalArticle

16 Citations (Scopus)

Abstract

The problem of deriving the particle size distribution directly from superparamagnetic magnetization curves is studied by three mathematical methods: (1) least-squares deviation with regularization procedure, (2) simulated annealing and (3) genetic algorithm. Software has been developed for the latest versions of all these methods and its performance compared for various models of underlying particle size distributions (Dirac δ-like, lognormal- and Gaussian-shaped). For single peak distributions all three methods give reasonable and similar results, but for bimodal distributions the genetic algorithm is the only acceptable one. The genetic algorithm is able to recover with the same precision both the lognormal and Gaussian single and double (mixed) model distributions. The sensitivity of the genetic algorithm - the most promising method - to uncertainty of measurements was also tested; correct peak position and its half width were recovered for Gaussian distributions, when the analysed data were contaminated with noise of up to 5% of MS.

Original languageEnglish
Pages (from-to)6027-6041
Number of pages15
JournalJournal of Physics A: Mathematical and General
Volume37
Issue number23
DOIs
Publication statusPublished - Jun 11 2004

Fingerprint

particle size distribution
Particle Size
genetic algorithms
Particle size analysis
Genetic algorithms
Genetic Algorithm
Bimodal
Gaussian distribution
simulated annealing
Mixed Model
Simulated annealing
normal density functions
Simulated Annealing
Magnetization
Paul Adrien Maurice Dirac
Least Squares
Regularization
Deviation
computer programs
deviation

ASJC Scopus subject areas

  • Physics and Astronomy(all)
  • Statistical and Nonlinear Physics
  • Mathematical Physics

Cite this

Langevin granulometry of the particle size distribution. / Kákay, Attila; Gutowski, M. W.; Takacs, L.; Franco, V.; Varga, L.

In: Journal of Physics A: Mathematical and General, Vol. 37, No. 23, 11.06.2004, p. 6027-6041.

Research output: Contribution to journalArticle

Kákay, Attila ; Gutowski, M. W. ; Takacs, L. ; Franco, V. ; Varga, L. / Langevin granulometry of the particle size distribution. In: Journal of Physics A: Mathematical and General. 2004 ; Vol. 37, No. 23. pp. 6027-6041.
@article{23aa63d7695c4650a897caff81a4f6cf,
title = "Langevin granulometry of the particle size distribution",
abstract = "The problem of deriving the particle size distribution directly from superparamagnetic magnetization curves is studied by three mathematical methods: (1) least-squares deviation with regularization procedure, (2) simulated annealing and (3) genetic algorithm. Software has been developed for the latest versions of all these methods and its performance compared for various models of underlying particle size distributions (Dirac δ-like, lognormal- and Gaussian-shaped). For single peak distributions all three methods give reasonable and similar results, but for bimodal distributions the genetic algorithm is the only acceptable one. The genetic algorithm is able to recover with the same precision both the lognormal and Gaussian single and double (mixed) model distributions. The sensitivity of the genetic algorithm - the most promising method - to uncertainty of measurements was also tested; correct peak position and its half width were recovered for Gaussian distributions, when the analysed data were contaminated with noise of up to 5{\%} of MS.",
author = "Attila K{\'a}kay and Gutowski, {M. W.} and L. Takacs and V. Franco and L. Varga",
year = "2004",
month = "6",
day = "11",
doi = "10.1088/0305-4470/37/23/005",
language = "English",
volume = "37",
pages = "6027--6041",
journal = "Journal Physics D: Applied Physics",
issn = "0022-3727",
publisher = "IOP Publishing Ltd.",
number = "23",

}

TY - JOUR

T1 - Langevin granulometry of the particle size distribution

AU - Kákay, Attila

AU - Gutowski, M. W.

AU - Takacs, L.

AU - Franco, V.

AU - Varga, L.

PY - 2004/6/11

Y1 - 2004/6/11

N2 - The problem of deriving the particle size distribution directly from superparamagnetic magnetization curves is studied by three mathematical methods: (1) least-squares deviation with regularization procedure, (2) simulated annealing and (3) genetic algorithm. Software has been developed for the latest versions of all these methods and its performance compared for various models of underlying particle size distributions (Dirac δ-like, lognormal- and Gaussian-shaped). For single peak distributions all three methods give reasonable and similar results, but for bimodal distributions the genetic algorithm is the only acceptable one. The genetic algorithm is able to recover with the same precision both the lognormal and Gaussian single and double (mixed) model distributions. The sensitivity of the genetic algorithm - the most promising method - to uncertainty of measurements was also tested; correct peak position and its half width were recovered for Gaussian distributions, when the analysed data were contaminated with noise of up to 5% of MS.

AB - The problem of deriving the particle size distribution directly from superparamagnetic magnetization curves is studied by three mathematical methods: (1) least-squares deviation with regularization procedure, (2) simulated annealing and (3) genetic algorithm. Software has been developed for the latest versions of all these methods and its performance compared for various models of underlying particle size distributions (Dirac δ-like, lognormal- and Gaussian-shaped). For single peak distributions all three methods give reasonable and similar results, but for bimodal distributions the genetic algorithm is the only acceptable one. The genetic algorithm is able to recover with the same precision both the lognormal and Gaussian single and double (mixed) model distributions. The sensitivity of the genetic algorithm - the most promising method - to uncertainty of measurements was also tested; correct peak position and its half width were recovered for Gaussian distributions, when the analysed data were contaminated with noise of up to 5% of MS.

UR - http://www.scopus.com/inward/record.url?scp=3042621444&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=3042621444&partnerID=8YFLogxK

U2 - 10.1088/0305-4470/37/23/005

DO - 10.1088/0305-4470/37/23/005

M3 - Article

VL - 37

SP - 6027

EP - 6041

JO - Journal Physics D: Applied Physics

JF - Journal Physics D: Applied Physics

SN - 0022-3727

IS - 23

ER -