pith. sign in

arxiv: 2605.17518 · v1 · pith:OWVZPWDYnew · submitted 2026-05-17 · ⚛️ physics.data-an · stat.AP· stat.ML

Integrating Bayesian Spectral Deconvolution and Expert Scientific Reasoning for Robust Peak Estimation

Hayato Okubo , Yoshifumi Amamoto , Toshimitsu Aritake , Hiroyuki Kumazoe , Shiryu Nakano , Evan Jamison , Satoshi Tanaka , Yoh-ichi Mototake This is my paper

Pith reviewed 2026-05-19 22:20 UTC · model grok-4.3

classification ⚛️ physics.data-an stat.APstat.ML
keywords Bayesian spectral deconvolutionGaussian process regressionpeak estimationphysical-property consistencyinfrared spectroscopymodel selectionpoly(lactic acid)expert scientific reasoning
0
0 comments X

The pith

Averaging physical-property likelihoods over Bayesian-inferred spectra selects models consistent with measured material properties.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes a Bayesian framework that couples spectral deconvolution with a physical-property regression layer implemented via Gaussian process regression. This layer evaluates candidate spectral structures by their consistency with independently measured physical-property values such as degradation rates. The central step averages the physical-property likelihood over posterior predictive spectra to select models that align inferred peaks with auxiliary data. The approach is shown to recover meaningful peaks in synthetic spectra with high noise or unknown backgrounds and in infrared spectra of poly(lactic acid) where spectrum-only methods fail. Readers would care because it formalizes the cross-referencing of spectral features with other experimental results that scientists perform when interpreting complex measurements.

Core claim

The authors claim that by averaging the physical-property likelihood over posterior predictive spectra inferred from Bayesian spectral deconvolution, the proposed method selects spectral models according to the consistency between inferred spectral structures and physical-property information. This enables recovery of physically meaningful peak structures, including weak peaks related to measured degradation rates in poly(lactic acid) IR spectra, that conventional Bayesian spectral deconvolution misses or misidentifies from spectra alone.

What carries the argument

The physical-property regression layer using Gaussian process regression, which supplies a consistency signal by relating posterior predictive spectra to independently measured physical properties and is coupled to Bayesian spectral deconvolution for model selection.

If this is right

  • Recovers physically meaningful peak structures from synthetic spectra containing high-intensity noise or unknown background components.
  • Identifies weak peaks in poly(lactic acid) infrared spectra that correspond to measured degradation rates.
  • Selects spectral models according to consistency with physical-property information beyond what spectrum features alone provide.
  • Remains reliable under conditions where conventional Bayesian spectral deconvolution alone fails or misidentifies peaks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The framework could apply to other spectroscopic methods such as Raman or NMR where auxiliary physical measurements are routinely available to constrain peak assignment.
  • Incorporating multiple physical properties at once in the regression layer might further reduce ambiguity in model selection for complex materials.
  • For high-throughput spectral analysis, the consistency check could reduce the volume of cases requiring manual expert review after automated deconvolution.

Load-bearing premise

The Gaussian process regression layer on physical properties supplies an independent and reliable consistency signal that correctly identifies physically meaningful peaks even when spectrum-only Bayesian deconvolution fails or misidentifies them.

What would settle it

Replacing the measured physical-property values with random uncorrelated numbers in the regression layer for the poly(lactic acid) spectra and checking whether the method then selects the same peaks as spectrum-only deconvolution; unchanged selection would indicate the consistency signal is not driving the result.

Figures

Figures reproduced from arXiv: 2605.17518 by Evan Jamison, Hayato Okubo, Hiroyuki Kumazoe, Satoshi Tanaka, Shiryu Nakano, Toshimitsu Aritake, Yoh-ichi Mototake, Yoshifumi Amamoto.

Figure 1
Figure 1. Figure 1: Graphical model of the proposed method. 𝑀 denotes the spectral model. For each sample 𝑗, the spectral parameter 𝜃 𝑗 defines both the posterior predictive spectrum 𝑌𝑗 and the observed spectrum 𝑌 obs 𝑗 through the spectral likelihood model. The physical property 𝑍 is modeled by Gaussian process regression using the latent function value 𝐹 conditioned on the spectral information. The red (blue) box indicates … view at source ↗
Figure 2
Figure 2. Figure 2: Noise-dominant synthetic spectral dataset used to validate the proposed method. Panels [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Example of fitting curves at the maximum a posteriori (MAP) solution for a synthetic [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Synthetic spectral dataset containing unknown background components. Panels (a)–(f) [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Synthetic spectral dataset prior to background subtraction. This figure displays the raw, [PITH_FULL_IMAGE:figures/full_fig_p020_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Example of fitting curves at the MAP solution for a synthetic spectrum ( [PITH_FULL_IMAGE:figures/full_fig_p021_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: IR spectral dataset of polylactic acid used as the real dataset. Each curve represents one [PITH_FULL_IMAGE:figures/full_fig_p023_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Results of MAP fitting for the IR spectrum of polylactic acid under the two candidate [PITH_FULL_IMAGE:figures/full_fig_p025_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Correspondence between ARD-based feature importance and spectra. Panels (a), (c), [PITH_FULL_IMAGE:figures/full_fig_p027_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Fitting results for artificial data 1 to 3. The solid blue line indicates the observed [PITH_FULL_IMAGE:figures/full_fig_p035_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Fitting results for artificial data 4 to 6. The solid blue line indicates the observed [PITH_FULL_IMAGE:figures/full_fig_p036_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: IR fitting results (Spectra 1–6). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p038_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: IR fitting results (Spectra 7–12). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p039_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: IR fitting results (Spectra 13–18). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p040_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: IR fitting results (Spectra 19–24). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p041_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: IR fitting results (Spectra 25–30). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p042_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: IR fitting results (Spectra 31–36). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p043_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: IR fitting results (Spectra 37–42). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p044_18.png] view at source ↗
Figure 19
Figure 19. Figure 19: IR fitting results (Spectra 43–48). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p045_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: IR fitting results (Spectra 49–54). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p046_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: IR fitting results (Spectra 55–60). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p047_21.png] view at source ↗
Figure 22
Figure 22. Figure 22: IR fitting results (Spectra 61–66). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p048_22.png] view at source ↗
Figure 23
Figure 23. Figure 23: IR fitting results (Spectra 67–68). The blue solid line represents the observed spectral [PITH_FULL_IMAGE:figures/full_fig_p049_23.png] view at source ↗
Figure 24
Figure 24. Figure 24: Spectra generated for different values of [PITH_FULL_IMAGE:figures/full_fig_p053_24.png] view at source ↗
Figure 25
Figure 25. Figure 25: Feature importance obtained by ARD. 2. Results of Applying the Proposed Method The parameters for each method in the proposed framework were set as described below to ensure clarity and consistency. A linear sum of the Gaussian functions was used as the spectral 53 [PITH_FULL_IMAGE:figures/full_fig_p053_25.png] view at source ↗
Figure 26
Figure 26. Figure 26: Top-ranked models and the FE at f 𝜎 = 0.01. model for Bayesian spectral deconvolution. The exchange Monte Carlo parameters in the spectral deconvolution model were defined as follows. The number of temperature points 𝐿 was set to 40, and based on the study by Nagata et al. [24], 𝛽𝑙 was set according to the following configuration. 𝛽𝑙 =    0.0 for 𝑙 = 1, 𝑑 𝑙−𝐿 for 𝑙 = 2, 3, . . . , 𝐿. (B5) The bur… view at source ↗
read the original abstract

Spectral deconvolution is essential for extracting peak structures that encode material properties and chemical structures, but conventional automated methods often fail when spectra contain high-intensity noise or unknown background components. In practice, scientists rarely interpret spectra in isolation. Instead, they identify physically meaningful peaks by relating spectral structures to auxiliary information such as physical-property values, chemical structures, and trends across related measurements. Here, we propose a Bayesian framework that integrates spectral deconvolution with a model of expert scientific reasoning. In this work, expert scientific reasoning refers to the practice of evaluating candidate spectral structures by their consistency with independently measured physical-property values, rather than to manual expert intervention during inference. We formalize this reasoning as a physical-property regression layer, implemented using Gaussian process regression, and couple it with Bayesian spectral deconvolution. By averaging the physical-property likelihood over posterior predictive spectra inferred from Bayesian spectral deconvolution, the proposed method selects spectral models according to the consistency between inferred spectral structures and physical-property information. We validate the framework using synthetic spectra with high-intensity noise or unknown backgrounds and infrared spectra of poly(lactic acid). The method recovers physically meaningful peak structures that conventional Bayesian spectral deconvolution misses or misidentifies from spectra alone, including weak peaks in poly(lactic acid) IR spectra related to measured degradation rates. These results demonstrate that integrating expert scientific reasoning with Bayesian spectral deconvolution enables robust peak estimation under conditions where spectrum-only inference is unreliable.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes a Bayesian framework that couples spectral deconvolution with a Gaussian process regression (GPR) layer modeling physical properties. By averaging the physical-property likelihood over posterior predictive spectra, the method selects spectral models according to consistency with independently measured auxiliary data such as degradation rates. Validation is reported on synthetic spectra with high noise or unknown backgrounds and on real poly(lactic acid) infrared spectra, where the integrated approach recovers weak peaks missed by spectrum-only Bayesian deconvolution.

Significance. If the central procedure holds, the work offers a principled way to incorporate auxiliary physical measurements into spectral model selection, potentially improving robustness in noisy or under-determined regimes common in materials characterization. The explicit use of posterior predictive averaging and the demonstration on both synthetic and experimental data constitute reproducible strengths that could influence Bayesian spectroscopy pipelines.

major comments (2)
  1. [Abstract and validation section] The central claim that averaging the physical-property likelihood over posterior predictive spectra yields correct recovery of peaks missed by spectrum-only deconvolution rests on the GPR layer supplying a reliable, non-redundant consistency signal. When the number of physical-property observations is modest or unmodeled confounders exist in the peak-to-property mapping, the GPR posterior may favor spurious alignments; this regime is not quantitatively bounded in the reported experiments.
  2. [Methods (physical-property regression layer)] The independence of the physical-property regression layer from the spectral data is asserted but not demonstrated via a controlled ablation (e.g., comparison of model selection with and without the GPR term when spectrum-only inference already fails). Without such a test, it remains unclear whether the reported improvement on poly(lactic acid) data is driven by the auxiliary signal or by implicit regularization.
minor comments (2)
  1. [Abstract] The phrase 'expert scientific reasoning' is defined in the abstract as evaluation against independently measured physical properties; this definition should be repeated verbatim in the introduction to avoid conflation with manual expert intervention.
  2. [Methods] Notation for the posterior predictive spectra and the averaged likelihood should be introduced with a single equation block rather than scattered references.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments, which have helped us identify areas for improvement in our manuscript. We address each of the major comments below and outline the revisions we will make.

read point-by-point responses

  1. Referee: [Abstract and validation section] The central claim that averaging the physical-property likelihood over posterior predictive spectra yields correct recovery of peaks missed by spectrum-only deconvolution rests on the GPR layer supplying a reliable, non-redundant consistency signal. When the number of physical-property observations is modest or unmodeled confounders exist in the peak-to-property mapping, the GPR posterior may favor spurious alignments; this regime is not quantitatively bounded in the reported experiments.

    Authors: We agree that the current experiments do not include quantitative bounds on the performance under modest numbers of physical-property observations or in the presence of unmodeled confounders. To address this, we will expand the validation section with additional experiments that systematically vary the number of auxiliary observations and introduce simulated confounders to delineate the reliable operating regime of the method. These additions will provide a more complete characterization of the conditions under which the integrated approach yields robust peak estimation. revision: yes

  2. Referee: [Methods (physical-property regression layer)] The independence of the physical-property regression layer from the spectral data is asserted but not demonstrated via a controlled ablation (e.g., comparison of model selection with and without the GPR term when spectrum-only inference already fails). Without such a test, it remains unclear whether the reported improvement on poly(lactic acid) data is driven by the auxiliary signal or by implicit regularization.

    Authors: The physical-property regression layer is based on auxiliary data collected independently of the spectra, such as degradation rates measured separately. Nevertheless, we concur that an explicit ablation study would better isolate the contribution of this layer. In the revised manuscript, we will include a controlled ablation comparing the full integrated model to the spectrum-only Bayesian deconvolution on both the synthetic datasets where spectrum-only inference fails and the poly(lactic acid) experimental data. This will clarify that the observed improvements stem from the consistency with auxiliary physical-property information. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation uses independent external physical-property measurements

full rationale

The central procedure infers posterior predictive spectra via Bayesian deconvolution then averages an external physical-property likelihood (GPR on independently measured values such as degradation rates) for model selection. This does not reduce by the paper's equations to a quantity defined solely in terms of its own fitted parameters or prior outputs. No self-citation load-bearing steps, uniqueness theorems, or ansatzes imported from the authors' prior work appear in the derivation chain. The method is validated against synthetic noisy spectra and real poly(lactic acid) IR data with external property labels, keeping the consistency signal non-redundant.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 1 invented entities

The approach rests on standard Bayesian spectral modeling plus a new regression layer whose hyperparameters are fitted; no new particles or forces are postulated.

free parameters (1)
  • Gaussian process hyperparameters
    Hyperparameters of the physical-property regression layer are fitted to data and influence the likelihood averaging step.
axioms (2)
  • standard math Standard Bayesian inference and posterior predictive sampling for spectral deconvolution
    The framework builds directly on established Bayesian spectral deconvolution methods.
  • domain assumption Gaussian process regression can capture the relationship between spectral structures and physical properties
    This modeling choice is introduced to represent consistency with auxiliary measurements.
invented entities (1)
  • physical-property regression layer no independent evidence
    purpose: To formalize expert scientific reasoning by scoring candidate spectra against independently measured properties
    This layer is newly introduced in the paper to integrate auxiliary information into model selection.

pith-pipeline@v0.9.0 · 5823 in / 1387 out tokens · 31158 ms · 2026-05-19T22:20:56.708512+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

45 extracted references · 45 canonical work pages

  1. [1]

    The regression of spectral data using the spectral deconvo- lution model yields information such as the number of peaks, peak positions, and peak variances [10]

    Spectral deconvolution model The spectral deconvolution model represents the spectral data as a linear combination of basis functions, such as Gaussian functions. The regression of spectral data using the spectral deconvo- lution model yields information such as the number of peaks, peak positions, and peak variances [10]. Consider the spectral data consi...

    work page

  2. [2]

    We employed Gaussian process regression as the Bayesian physical-property regression model in this study

    Physical-property regression model The physical-property regression model connects the spectral data to the corresponding physi- cal property values. We employed Gaussian process regression as the Bayesian physical-property regression model in this study. In the physical-property regression model, a single sample consti- tutes a pair of spectral data𝑌:={𝑦...

    work page

  3. [3]

    The simplest physical prior knowledge is the value of a property related to a spectrum

    Model of spectral deconvolution process followed by scientists We modeled the method followed by scientists for extracting high-precision information by analyzing the spectral data in combination with physical prior knowledge. The simplest physical prior knowledge is the value of a property related to a spectrum. Given the spectral data and corresponding ...

    work page

  4. [4]

    Marginalization of spectral deconvolution model The method for sampling from the posterior of the spectral deconvolution model using Replica Exchange Monte Carlo is described here. The purpose of the marginalization calculation for the spectral deconvolution model, ∫ 𝑑𝜃 𝑝(Y |𝜗)p(Y obs|𝜗)𝑝(𝜗|𝑀), is to compute a sample set from the predictive distribution𝑝(...

    work page

  5. [5]

    Marginalization of Physical-Property Regression Models This section describes the marginalization calculation for the physical-property regression model defined in Sec. II B 2. The spectral samples{𝑌 𝑘 }𝑁samp 𝑘=1 drawn from𝑝(𝑌|𝜗)are combined with the measurement data𝑍to form the dataset{Y 𝑘 , 𝑍} 𝑁𝑠𝑎𝑚 𝑝 𝑘=1 .𝑁 𝑠𝑎𝑚 𝑝 denotes the sample size drawn from the p...

    work page

  6. [6]

    This is shown in the following equation

    Marginalization of models of spectral deconvolution processes used by scientists Marginalization overYmeans taking the expectation (average value) of𝑝(𝑍|Y)with respect to the probability distribution𝑝(Y |Y obs, 𝑀), whereYis the set of all possible values,Y obs represents the observed values, and𝑀denotes the model. This is shown in the following equation. ...

    work page 2074

  7. [7]

    A. G. Harrison,Chemical ionization mass spectrometry(Routledge, 2018)

    work page 2018

  8. [8]

    R. Fan, X. Yang, C. F. Drury, and Z. Zhang, Curve-fitting techniques improve the mid-infrared analysis of soil organic carbon: a case study for brookston clay loam particle-size fractions, Scientific Reports 8, 12174 (2018)

    work page 2018

  9. [9]

    Dazzi and C

    A. Dazzi and C. B. Prater, AFM-IR: Technology and applications in nanoscale infrared spectroscopy and chemical imaging, Chemical Reviews117, 5146 (2017), pMID: 27958707, https://doi.org/10.1021/acs.chemrev.6b00448

    work page doi:10.1021/acs.chemrev.6b00448 2017

  10. [10]

    Kneipp, H

    K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari, and M. S. Feld, Ultrasensitive chemical analysis by Raman spectroscopy, Chemical reviews99, 2957 (1999)

    work page 1999

  11. [11]

    A. E. Derome,Modern NMR techniques for chemistry research(Elsevier, 2013)

    work page 2013

  12. [12]

    C. F. Holder and R. E. Schaak, Tutorial on powder X-ray diffraction for characterizing nanoscale materials, ACS Nano13, 7359 (2019), pMID: 31336433

    work page 2019

  13. [13]

    Buglioni, F

    L. Buglioni, F. Raymenants, A. Slattery, S. D. A. Zondag, and T. No ¨el, Technological innovations in photochemistry for organic synthesis: Flow chemistry, high-throughput experimentation, scale-up, and photoelectrochemistry, Chemical Reviews122, 2752 (2022), pMID: 34375082

    work page 2022

  14. [14]

    D. R. Baer, K. Artyushkova, C. R. Brundle, J. E. Castle, M. H. Engelhard, K. J. Gaskell, J. T. Grant, R. T. Haasch, M. R. Linford, C. J. Powell, A. G. Shard, P. M. A. Sherwood, and V. S. Smentkowski, Practical Guides for X-Ray Photoelectron Spectroscopy (XPS): First Steps in planning, conducting and reporting XPS measurements, Journal of Vacuum Science & ...

    work page doi:10.1116/1.5065501 2019

  15. [15]

    J. Near, A. Harris, C. Juchem, R. Kreis, M. Marja ´nska, G. ¨Oz, J. Slotboom, M. Wilson, and C. Gas- parovic, Preprocessing, analysis and quantification in single-voxel magnetic resonance spectroscopy: Experts’ consensus recommendations, NMR in Biomedicine34, e4257 (2021)

    work page 2021

  16. [16]

    M. O. Kenji Nagata, Seiji Sugita, Bayesian spectral deconvolution with the exchange Monte Carlo method, Neural Networks28, 82 (2012)

    work page 2012

  17. [17]

    Fasano, P

    A. Fasano, P. Ade, M. Aravena, E. Barria, A. Beelen, A. Benoit, M. B´ethermin, J. Bounmy, O. Bourrion, G. Bres, M. Calvo, A. Catalano, C. D. Breuck, F.-X. D ´esert, C. Dubois, C. Dur ´an, T. Fenouillet, 30 J. Garcia, G. Garde, J. Goupy, C. Hoarau, W. Hu, G. Lagache, J.-C. Lambert, F. Levy-Bertrand, A. Lundgren, J.-F. Mac ´ıas-P´erez, J. Marpaud, A. Monfar...

    work page arXiv 2024

  18. [18]

    C. Wang, L. Xiao, C. Dai, A. H. Nguyen, L. E. Littlepage, Z. D. Schultz, and J. Li, A statistical approach of background removal and spectrum identification for SERS data, Scientific Reports10, 1460 (2020)

    work page 2020

  19. [19]

    Diehl, J

    J. Diehl, J. Knollm¨ uller, and O. Schulz, Bias-free estimation of signals on top of unknown backgrounds, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detec- tors and Associated Equipment1063, 169259 (2024)

    work page 2024

  20. [20]

    Macaro and R

    C. Macaro and R. Prado, Spectral decompositions of multiple time series: a Bayesian non- parametric approach, Psychometrika79, 105 (2014), research Support, U.S. Gov’t, Non-P.H.S., https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3925306/

    work page 2014

  21. [21]

    C. C. Giac and T. V. Thanh, Establishing a spectroscopic analysis procedure for identifying the molecular structure of organic compounds to enhance chemistry students’ problem-solving skills, World Journal of Chemical Education13, 1 (2025)

    work page 2025

  22. [22]

    Yui, Fusion of spectroscopy and geology: Earth’s interior environment revealed 5 through light, Japanese Magazine of Mineralogical and Petrological Sciences52, 230110a (2023)

    K. Yui, Fusion of spectroscopy and geology: Earth’s interior environment revealed 5 through light, Japanese Magazine of Mineralogical and Petrological Sciences52, 230110a (2023)

    work page 2023

  23. [23]

    Kashiwamura, S

    S. Kashiwamura, S. Katakami, R. Yamagami, K. Iwamitsu, H. Kumazoe, K. Nagata, T. Okajima, I. Akai, and M. Okada, Bayesian spectral deconvolution of X-ray absorption near edge structure discriminating between high- and low-energy domains, Journal of the Physical Society of Japan91, 074009 (2022)

    work page 2022

  24. [24]

    Han and R

    N. Han and R. J. Ram, Bayesian modeling and computation for analyte quantification in complex mixtures using Raman spectroscopy, Computational Statistics & Data Analysis143, 106846 (2020)

    work page 2020

  25. [25]

    C. M. Bishop,Pattern Recognition and Machine Learning (Information Science and Statistics) (Springer-Verlag, Berlin, Heidelberg, 2006)

    work page 2006

  26. [26]

    W. K. Hastings, Monte Carlo sampling methods using Markov chains and their applications, Biometrika 57, 97 (1970)

    work page 1970

  27. [27]

    Latuszy ´nski, M

    K. Latuszy ´nski, M. T. Moores, and T. Stumpf-F ´etizon, MCMC for multi-modal distributions, arXiv preprint arXiv:2501.05908 (2025). 31

    work page arXiv 2025

  28. [28]

    Marinari and G

    E. Marinari and G. Parisi, Simulated tempering: A new Monte Carlo scheme, Europhysics Letters (EPL)19, 451–458 (1992)

    work page 1992

  29. [29]

    IBA, Extended ensemble Monte Carlo, International Journal of Modern Physics C12, 623–656 (2001)

    Y. IBA, Extended ensemble Monte Carlo, International Journal of Modern Physics C12, 623–656 (2001)

    work page 2001

  30. [30]

    Nagata and S

    K. Nagata and S. Watanabe, Asymptotic behavior of exchange ratio in exchange Monte Carlo method, Neural Networks21, 980 (2008)

    work page 2008

  31. [31]

    P. H. C. Eilers and H. F. Boelens,Baseline Correction with Asymmetric Least Squares Smoothing, Tech. Rep. (Leiden University Medical Centre Report, 2005)

    work page 2005

  32. [32]

    L. P. Malone, S. M. Best, and R. E. Cameron, Accelerated degradation testing impacts the degradation processes in 3D printed amorphous PLLA, Frontiers in Bioengineering and Biotechnology12, 1419654 (2024)

    work page 2024

  33. [33]

    Meaurio, N

    E. Meaurio, N. L ´opez-Rodr´ıguez, and J.-R. Sarasua, Infrared spectrum of poly(l-lactide): application to crystallinity studies, Macromolecules39, 9291 (2006)

    work page 2006

  34. [34]

    Takahashi, Y

    K. Takahashi, Y. Amamoto, H. Kikutake, M. Ito, A. Takahara, and T. Onishi, Random forest analysis of X-ray diffraction and scattering data on crystalline polymer, Journal of Computer Chemistry, Japan 20, 103 (2021), (in Japanese)

    work page 2021

  35. [35]

    Zhang, H

    J. Zhang, H. Tsuji, I. Noda, and Y. Ozaki, Weak intermolecular interactions during the melt crystalliza- tion of poly(l-lactide) investigated by two-dimensional infrared correlation spectroscopy, The Journal of Physical Chemistry B108, 11514 (2004)

    work page 2004

  36. [36]

    Kister, G

    G. Kister, G. Cassanas, and M. Vert, Effects of morphology, conformation and configuration on the IR and Raman spectra of various poly(lactic acid)s, Polymer39, 267 (1998)

    work page 1998

  37. [37]

    Zhang, S

    Z.-M. Zhang, S. Chen, and Y.-Z. Liang, Baseline correction using adaptive iteratively reweighted penalized least squares, Analyst135, 1138 (2010)

    work page 2010

  38. [38]

    Meyns, F

    M. Meyns, F. Dietz, C.-S. Weinhold, H. Z¨ uge, S. Finckh, and G. Gerdts, Multi-feature round silicon membrane filters enable fractionation and analysis of small micro- and nanoplastics with Raman spectroscopy and nano-FTIR, Analytical Methods15, 606 (2023)

    work page 2023

  39. [39]

    Tokuda, K

    S. Tokuda, K. Nagata, and M. Okada, Simultaneous estimation of noise variance and number of peaks in Bayesian spectral deconvolution, Journal of the Physical Society of Japan86, 024001 (2017)

    work page 2017

  40. [40]

    Mototake, M

    Y.-i. Mototake, M. Mizumaki, I. Akai, and M. Okada, Bayesian hamiltonian selection in X-ray photo- electron spectroscopy, Journal of the Physical Society of Japan88, 034004 (2019). 32

    work page 2019

  41. [41]

    Kotani, H

    A. Kotani, H. Mizuta, T. Jo, and J. Parlebas, Theory of core photoemission spectra in CeO2, Solid State Communications53, 805 (1985). 33 Appendix A: Fitting results of this study This section presents the detailed fitting curves for all samples of the artificial data and polylactic acid IR spectral data discussed in Section 3. Specifically, we compare the...

    work page 1985

  42. [42]

    Fitting results for artificial data This subsection presents the fitting results for all six artificial spectra generated for different peak-to-peak distancesΔ. The figure shows that regardless of the value ofΔ, that is, the degree of peak overlap, the proposed method (orange dashed line) accurately captures the true two-peak structure underlying the obse...

    work page

  43. [43]

    IR Spectrum Fitting Results This subsection lists the fitting results for each model applied to the IR spectra of polylactic acid obtained under 68 conditions with different crystallization temperatures and concentrations (Figs. 12–23). These results indicate that the proposed method stably extracts two weak peaks in the biodegradability-related region. 3...

    work page

  44. [44]

    Here, we employed the spectroscopic model proposed by Kotani et al

    Verification Dataset We used a dataset containing artificially generated inner-shell XPS spectral data generated from an XPS spectroscopic model and the corresponding physical parameters related to the electronic levels specified in the data-generating process [34]. Here, we employed the spectroscopic model proposed by Kotani et al. [35] for 4f-orbital el...

    work page

  45. [45]

    A linear sum of the Gaussian functions was used as the spectral 53 Figure 26: Top-ranked models and the fFE at𝜎=0.01

    Results of Applying the Proposed Method The parameters for each method in the proposed framework were set as described below to ensure clarity and consistency. A linear sum of the Gaussian functions was used as the spectral 53 Figure 26: Top-ranked models and the fFE at𝜎=0.01. model for Bayesian spectral deconvolution. The exchange Monte Carlo parameters ...

    work page