資源描述：
《Maximum Entropy Deconvolution Appl. Spectros》由會(huì)員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、MaximumEntropyDeconvolutionofInfraredSpectra:UseofaNovelEntropyExpressionWithoutSignRestrictionVI?CTORA.LO?RENZ-FONFRI?A*andESTEVEPADRO?SUnitatdeBiof??sica.DepartamentdeBioqu??micaideBiologiaMolecular,FacultatdeMedicina,UniversitatAutoánomadeBarcelona,08193Bellaterr
2、a,Barcelona,SpainAbsorbanceanddifferenceinfraredspectraareoftenacquiredaim-plexLaplacetransform,3linearpredictionbyautoregres-ingtocharacterizeproteinstructureandstructuralchangesofpro-sivemodeling,4±6wavelet-transform,7,8anddeconvolu-teinsuponligandbinding,aswellas
3、formanyotherchemicalandtion.biochemicalstudies.Theiranalysisrequiresasa?rststeptheiden-Deconvolutionhasbecomeoneofthemostusedmeth-ti?cationofthecomponentbands(number,position,andarea)andodsfornarrowingpurposes.Deconvolutionisanunstableasasecondsteptheirassignment.Th
4、e?rststepoftheanalysisisprocessanddifferentstrategiesforthestabilizationofthechallengedbythehabituallystrongbandoverlapininfraredspec-tra.Therefore,itisusefultomakeuseofamathematicalmethoddeconvolutedsolutionhavebeenproposed,namely,theuseofasmoothing?lterinFourierde
5、convolution;1,9±12abletonarrowthecomponentbandstotheextenttoeliminate,oratleastreduce,thebandoverlap.Additionally,tobeofgeneralthecombinationofdeconvolutionandlinearprediction;13±16applicabilitythismethodshouldpermitnegativevaluesfortheso-thecombinationofFourierdeco
6、nvolutionandwaveletlution.Wepresentamaximumentropydeconvolutionapproachfortransform;17,18iterativedeconvolutionwithconstraints,theband-narrowingofabsorbanceanddifferencespectrashowingalsoknownasmaximumlikelihoodrestoration;19±21andtherequiredcharacteristics,whichuse
7、sthegeneralizednegativetheapplicationofinversiontheory,inwhichweincludeBurg-entropy(Itakura±Sa?ètodiscrepancy)generalizedfordiffer-19,22relatedmethodslikeregularizationandprobabilisticencespectra.Wepresentresultsonsyntheticnoisyabsorbanceand(Bayesian)methods.23diffe
8、rencespectra,aswellasonexperimentalinfraredspectrafromthemembraneproteinbacteriorhodopsin.Regularizedandprobabilisticmethodsconsiderthede-
