Easily Computed Marginal Likelihoods for Multivariate Mixture Models Using the THAMES Estimator

Adrian E. Raftery; Marie Perrot-Dock\`es; Martin Metodiev; Nicholas J. Irons; Pierre Latouche

REVIEW 4 major objections 5 minor 63 references

Easily Computed Marginal Likelihoods for Multivariate Mixture Models Using the THAMES Estimator

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper adapts the truncated harmonic mean estimator to multivariate mixture models by symmetrising over component labels and making the permutation sum tractable through a QDA-based ordering constraint.

desk verdict The stress-test concern about the estimator's numerator/denominator is a misreading; Equations (1) and (5) are the standard reciprocal form, and the paper's central idea is sound, though the proofs leave the estimated truncation region and MCMC dependence unquantified. read the letter →

arxiv 2504.21812 v3 pith:5IOZXLY4 submitted 2025-04-30 stat.ME

Martin Metodiev , Nicholas J. Irons , Marie Perrot-Dockès , Pierre Latouche , Adrian E. Raftery This is my paper · ORCID

classification stat.ME MSC 62F1562-0462F12

keywords Bayesianstatisticsbridgesamplingharmonicmeanestimatorlabelswitchingmarginallikelihoodestimationmixturemodellingtruncatedoverlapgraphs

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that a symmetric version of the truncated harmonic mean estimator (THAMES) computes the marginal likelihood of a multivariate mixture model from nothing more than the unnormalised log-posterior and a relabelled posterior parameter sample. The estimator is claimed to be consistent, asymptotically normal and of finite variance, and invariant to label switching, so no simulation from hidden allocation vectors or randomly permuted samples is needed. The naive definition averages over all $G!$ label permutations, which would be intractable, but the paper shows that the sum reduces to a small set of topological orderings defined by a quadratic-discriminant-analysis-based ordering score, so the cost stays manageable even for $G=15$. In simulations with analytically known marginal likelihoods the estimator converges to the truth while bridge sampling can be off by a factor of $G!$, and on real datasets it selects plausible numbers of components. If the claims hold, applied Bayesians obtain a simple and generic tool for model choice in high-dimensional mixture models.

What carries the argument

The load-bearing object is the symmetric THAMES estimator of Equation (5), viewed as a reciprocal importance sampling estimator whose proposal is a mixture of uniform distributions on the symmetrised truncation set. Its tractability comes from the partial order defined by the ordering score $W(\xi_g) = \hat{g}(\xi_g|I(G)) + 1 - \hat{w}_{\hat{g}(\xi_g|I(G))}(\xi_g|I(G))$, built from quadratic discriminant analysis restricted to a maximum independent set of non-overlapping components; the inequalities $W(\xi_{g_1}) < W(\xi_{g_2})$ are verified on the ellipsoid $E_{\hat{\theta},\hat{\Sigma},c}$, producing an adjacency matrix whose topological orderings are exactly the nonzero terms of the estimator. The graph machinery also yields the criterion of overlap, a separate estimate of the number of distinguishable components. In short, the ordering constraint converts an exponential sum over label permutations into a small sum over graph-theoretic orderings without changing the value of the estimator.

What would settle it

On a small mixture with an exactly computable marginal likelihood, take one long MCMC run and compute THAMES on the raw chain, on a chain thinned by a factor of 10, and on a set of independent draws; if the three estimates differ by more than their reported Monte Carlo standard errors, the independence and split-sample assumptions are violated in a way that matters.

Watch

Extended reading notes

Core claim

The central discovery is that the mixture-adapted THAMES, defined as a uniform average over label permutations of a truncated reciprocal-importance-sampling estimator on the set $B_{\hat{\theta},\hat{\Sigma},c,\alpha} = E_{\hat{\theta},\hat{\Sigma},c} \cap \{\theta : \pi(\theta)L(\theta) > \hat{q}_\alpha\}$, is both theoretically sound and computationally tractable. The paper proves unbiasedness on the reciprocal scale, consistency, asymptotic normality and finite variance, and shows that this symmetrisation is optimal among mixtures of truncated harmonic mean estimators. The computational breakthrough is Equation (8): because the ordering constraint $W(\xi_1) \le \cdots \le W(\xi_G)$ holds throughout the truncation region for most pairs of components, the sum over $G!$ permutations equals a sum over the topological orderings of a directed graph, and only those orderings need to be evaluated. The score $W$ is derived from quadratic discriminant analysis on the relabelled posterior sample, restricted to a maximum independent set of non-overlapping components found via overlap graphs. An additional identity lets the estimator handle empty components by reducing to a lower-dimensional model, which matters when many components are fitted and some capture no data points.

Load-bearing premise

The theoretical guarantees assume the MCMC draws used in the second half are effectively independent of each other and of the first half, which defines the truncation set; the paper states this is approximate, and the proofs do not quantify the effect of dependence or of estimating the truncation set from the data.

Editorial extensions

If this is right

Marginal likelihood becomes a routine post-processing step for multivariate Gaussian mixtures with up to at least 15 components, using only relabelled posterior draws and unnormalised log-posterior values.
Because hidden allocation vectors are not required, the estimator works with samplers that do not produce them, widening the class of tractable mixture models.
Model selection for mixtures can be performed with an estimator that is a formal reciprocal importance sampler, so its consistency and asymptotic normality bring the usual machinery of Monte Carlo error assessment.
In well-separated settings the $G!$-factor bias seen in bridge sampling is avoided, so the two estimators can serve as mutual checks in practice.
The criterion of overlap provides a complementary answer to a different question: how many distinguishable clusters exist, alongside the marginal likelihood's answer to how many components the model needs.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the estimator only needs a relabelled posterior and the log posterior, a natural extension is to models beyond Gaussian mixtures, such as t-mixtures or mixtures of experts; the ordering-score construction would need re-testing when the relabelled posterior is not approximately normal.
The split-sample independence assumption could be probed directly: on a long chain, compare THAMES computed from heavily thinned, lightly thinned, and independent draws; the difference in estimates is a measure of how much the proofs rely on the approximation.
The $G!$ discrepancy with bridge sampling in the well-separated simulation suggests a diagnostic for practitioners: when two marginal likelihood estimators disagree by a large factor, test whether one proposal is label-asymmetric, since the paper's symmetrisation is what removes that bias.
For very large numbers of components, the paper's own discussion indicates variance grows as the ellipsoid must be shrunk to keep the ordering set small; a useful extension would be an automated rule for trading off ellipsoid size against variance.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

The stress-test note does not hold up on close reading. Equations (1) and (5) put πL in the denominator: the expression is (1/V(B)) / (π(θ)L(θ)), not πL in the numerator. I checked the main text and Supplement A.2; they are consistent with each other and with the usual reciprocal importance sampling identity. So the core definition is fine.

What is actually new here is a useful combination: symmetrisation over label permutations (borrowed from Berkhof) plus Reichl's ellipsoid-HPD truncation, made computationally tractable via the overlap-graph ordering and QDA-based constraint. The CO criterion is a genuinely new auxiliary tool for choosing the number of distinct components. The univariate simulation, with analytically known marginal likelihoods across 50 datasets and three misspecification scenarios, is solid evidence: THAMES tracks the truth where bridge sampling and the Reichl estimator go off. The multivariate evidence is thinner — one well-separated artificial dataset per setting — but it does show the method scales to G=15, and the real-data examples are sensible.

The soft spots are real but not fatal. The consistency and variance proofs treat B and its volume as fixed, while the implementation estimates them from the data; the split-sample construction is approximate, and the independence assumption in the optimality theorem is acknowledged but not quantified. The QDA-based ordering relies on approximate normality, which is reasonable for well-identified mixtures but not guaranteed. The 'arbitrarily high number of components' claim in the abstract is hyperbolic; the paper's own discussion notes that high overlap forces a smaller ellipsoid and higher variance. These are addressable with sharper assumptions or more extensive simulations, and they do not undermine the central estimator.

The paper is honest, clearly written, and ships an R package. It is a methodological contribution that applied Bayesians would find genuinely useful for model selection in multivariate mixtures. I would send it to peer review; a good referee should push on the dependence and estimated-truncation issues, but the work deserves the time.

Referee Report

4 major / 5 minor

Summary. This paper proposes a symmetrized version of the truncated harmonic mean estimator (THAMES) for the marginal likelihood of mixture models. The estimator averages a Reichl-type volume-truncated harmonic mean over all G! label permutations of relabelled posterior draws, and uses an ordering constraint based on quadratic discriminant analysis and overlap graphs to reduce the cost of the permutation sum. The authors prove in a supplement symmetry, unbiasedness, consistency, asymptotic normality, and a variance-optimality property for an idealized RIS formulation, provide an empty-component correction based on Nobile's identity, and evaluate the estimator on simulated and real datasets, including comparisons with bridge sampling and Reichl's estimator. The typeset estimator in Eq. (5) has π(θ)L(θ) in the denominator, matching the RIS derivation in Supplement A.2; the alleged reciprocal inconsistency does not survive a close reading.

Significance. If the theoretical claims were established for the actual algorithm, this would be a valuable addition to the toolkit for model selection in mixture models: it is generic, uses only relabelled posterior draws, and the reported computations appear feasible for G=15. The paper includes useful analytic benchmarks (Theorem 5.6 and Supplement D), a detailed supplement with proofs, and an R package. The central gap is that the implemented estimator differs from the one analyzed: the truncation set, its volume, and the ordering constraint are estimated from the same MCMC sample used in the average, so the stated guarantees are not proven for the actual procedure. The work is therefore promising but requires a substantial revision to close the gap between the ideal estimator and the implementation.

major comments (4)

[Section 3.2; Supplement A.2] The RIS analysis in Supplement A.2 treats h(θ) = (1/G!)∑_o 1_B(P_o θ)/V(B) as fixed and independent of the sample used in the average. In the implementation, the quadratic discriminant classifier (using component means and covariances estimated on the second half), the ordering set Ω, the threshold q̂_α, and the Monte Carlo volume V̂(B) are all estimated from the same MCMC output. The first/second-half split covers only θ̂ and Σ̂. Hence the estimator actually computed is not the RIS estimator whose unbiasedness and CLT are proved; the data-adaptive construction requires either an independent evaluation sample or a separate theorem quantifying the effect of estimating these inputs.
[Eq. (3) and Eq. (5)] Equation (5) is written with the exact volume V(B), but the algorithm replaces it by the Monte Carlo estimate from Eq. (3) with N=T. The proofs in Supplement A.2 use the exact volume. Because E[1/V̂] does not equal 1/V in general, the finite-sample unbiasedness claim fails for the implemented ratio estimator, and consistency of the ratio needs a separate argument, for example uniform convergence of V̂ and independence from the inner average.
[Supplement A.1 and A.2] The variance-optimality theorems (Theorem 5.1 in Supplement A.1 and Theorem 5.3 in Supplement A.3) explicitly assume independent posterior draws, and the main text acknowledges this is approximate. Supplement A.2 states that the symmetry, unbiasedness, consistency, and asymptotic normality results make no such assumptions, but it does not state the Markov-chain conditions used. Since the headline claims are for MCMC samples, the paper should state explicit ergodicity or mixing conditions and clarify that 'unbiased' is an iid-sample or asymptotic statement.
[Theorem 5.4 / Eq. (8)] The identity replacing the G! permutation sum by the sum over topological orderings Ω is exact only when the adjacency matrix Δ is known exactly. The algorithm estimates Δ by checking W-inequalities on a finite Monte Carlo sample from the ellipsoid, as described in Supplement B.4; a misclassification changes the value of the estimator, not merely its computational cost. No error bound or consistency statement is given for this approximation.

minor comments (5)

[Eq. (1) and Eq. (5)] The typesetting of Equations (1) and (5) should make the reciprocal explicit, for example by writing 1/(V(B)π(θ)L(θ)), since the current line break invites misreading as π(θ)L(θ) in the numerator.
[Section 3.2] The text refers to 'Equation (17)' when discussing computation of the THAMES, but in the main text the relevant display is Equation (8); Equation (17) is in the supplement.
[Supplement A.2] The statement that the posterior density is bounded below whenever h is not 0 should explicitly invoke the threshold q̂_α > 0 on B, rather than relying on the boundedness of the support alone.
[Figures 2 and 8] The captions of Figures 2 and 8 describe counting points in the ellipsoid, but the relevant condition is membership in B = E ∩ {πL > q̂_α}; please clarify the captions.
[Figure 4] The comparison with bridge sampling would be more informative if the paper specified whether bridge sampling was applied to the relabelled or the raw posterior sample, since non-symmetric estimators applied to relabelled draws are not automatically estimators of the same quantity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the mixture THAMES construction and its empirical checks are self-contained; the printed Eq (5) versus Supplement A.2 proof mismatch is a correctness defect, not a fit or self-citation reduction.

full rationale

I find no circular step in the paper's derivation chain. The symmetric THAMES is defined as an average over permutations of a truncated estimator, and the claimed consistency, asymptotic normality, and finite variance are supported by the classical reciprocal importance sampling identity in Supplement A.2, not by fitting a parameter to the benchmark marginal likelihoods. The ordering reduction in Supplement A.4 is an exact sum identity over all permutations, and the empty-component recursion follows Nobile's identity with a separate consistency assumption on the lower-dimensional estimator, so it is not a circular reuse of the target quantity. Tuning parameters are chosen by heuristics or by a chi-squared goodness-of-fit criterion, and the empirical evaluation compares against analytic marginal likelihoods (Theorem 5.6, Supplement D) and external estimators such as bridge sampling and Reichl's estimator, so the validation is not internally calibrated to the answer. The paper does cite the authors' original THAMES paper [41] for the ellipsoid constant and for an optimality result under normal posteriors, but that cited work is a separate published method with its own assumptions, and the mixture-specific extension is independently constructed and tested here; those self-citations are not load-bearing for the central claim. One caveat, not a circularity: as printed, Equations (1) and (5) put pi(theta)L(theta) in the numerator, whereas Supplement A.2's RIS derivation requires the reciprocal 1/(pi(theta)L(theta)) in the summand. This is an internal inconsistency or typographical defect that would make the proof not apply to the displayed estimator, and it should be corrected before the theoretical claims are accepted; however, it is a proof-consistency issue rather than a reduction of the result to its own inputs.

Assumptions & free parameters 5 free parameters · 5 assumptions · 3 invented entities

The estimator rests on standard Bayesian and Monte Carlo machinery plus several stated assumptions: exchangeable priors (symmetric posterior), approximately i.i.d. posterior samples, bounded posterior density on the truncation set, and approximate normality for the QDA ordering trick. The tuning parameters c, alpha, and the volume Monte Carlo size N are heuristics selected from the posterior sample or from prior work, not fitted to the target marginal likelihood. The QDA ordering constraint and the criterion of overlap are new constructs introduced by this paper.

free parameters (5)

c (ellipsoid radius) = sqrt(R+1)
Set following Metodiev et al. [41] as an asymptotically optimal radius for normal posteriors; not fitted to the target marginal likelihood.
alpha (HPD tuning parameter) = Chosen by minimizing Kolmogorov distance to a shifted/scaled chi-square distribution, capped at 0.5
A heuristic based on the posterior sample; the paper notes alpha=0.5 works well. Not fitted to the benchmark.
N (volume Monte Carlo size) = N = T
Number of uniform draws in the ellipsoid used to estimate the volume of B; set equal to the posterior sample size.
Omega bound = 50,000
If the estimated number of topological orderings exceeds 50,000, the ellipse is shrunk; a computational heuristic.
p0 threshold = 1/T
If the estimated probability of an empty component exceeds 1/T, the model dimension is reduced via the Nobile identity; an implementation threshold.

assumptions (5)

domain assumption The prior distribution of (tau_g, xi_g) does not depend on the index g, resulting in a symmetric posterior.
Stated in Section 3.1. Without exchangeability, the label-switching framework and the G! symmetrisation do not apply.
domain assumption The MCMC sample after burn-in and thinning is an approximately i.i.d. sample from the posterior.
Assumed in Supplement A.1 and A.2 for unbiasedness, consistency, and asymptotic normality. The paper acknowledges it is approximate.
ad hoc to paper The relabelled posterior distribution is approximately Gaussian for QDA to provide a reliable ordering constraint.
Section 3.2 states the condition needed to apply QDA is asymptotic normality. This is used to justify the ordering constraint's efficiency, not the estimator's consistency.
domain assumption The posterior density is bounded below on the truncation set B, ensuring finite RIS variance.
Supplement A.2, the condition for Equation (15). Holds if B is contained in an HPD region where the unnormalized posterior exceeds a positive threshold.
domain assumption For the Nobile identity (Theorem 5.5), component parameters are a priori independent and their prior does not depend on the total number of components G.
Stated in Theorem 5.5 and noted in Section 3.3. Not used for the univariate datasets.

invented entities (3)

Symmetric THAMES for mixture models (Equation 5) independent evidence
purpose: Marginal likelihood estimator invariant to label switching
The estimator is the paper's main result, tested on simulations with known marginal likelihoods and on real datasets.
Overlap graphs and the criterion of overlap (CO) independent evidence
purpose: Visualize component overlap and estimate the number of distinct components
Validated on galaxy, enzyme, acidity, Swiss banknote, and BUPA datasets, where CO gives interpretable cluster numbers.
QDA-based ordering constraint W
purpose: Reduce the G! permutation sum to topological orderings, enabling efficient computation
A heuristic device with no formal optimality guarantee, demonstrated to work in simulations. If it fails, the ellipse is shrunk, which affects variance but not the estimator's defining equation.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Easily Computed Marginal Likelihoods for Multivariate Mixture Models Using the THAMES Estimator." pith.science (2026). https://pith.science/paper/5IOZXLY4

@misc{pith2026250421812,
  author       = {Pith},
  title        = {Pith review of: Easily Computed Marginal Likelihoods for Multivariate Mixture Models Using the THAMES Estimator},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5IOZXLY4}},
  note         = {Machine review of arXiv:2504.21812}
}

read the original abstract

We present a new version of the truncated harmonic mean estimator (THAMES) for univariate or multivariate mixture models. The estimator computes the marginal likelihood from Markov chain Monte Carlo (MCMC) samples, is consistent, asymptotically normal and of finite variance. In addition, it is invariant to label switching, does not require posterior samples from hidden allocation vectors, and is easily approximated, even for an arbitrarily high number of components. Its computational efficiency is based on an asymptotically optimal ordering of the parameter space, which can in turn be used to provide useful visualisations. We test it in simulation settings where the true marginal likelihood is available analytically. It performs well against state-of-the-art competitors, even in multivariate settings with a high number of components. We demonstrate its utility for inference and model selection on univariate and multivariate data sets.

Figures

Figures reproduced from arXiv: 2504.21812 by the authors.

**Figure 2.** A toy example to illustrate the computation of the THAMES; the points that [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

**Figure 3.** Boxplots of the error log(Zˆ) − log(Z) of bridge sampling, the estimator from Reichl [51], and the THAMES for 50 different simulations of the data for increasing levels of overlap ρ = 0, 1/2, 1 and 3 different scenarios corresponding to the true model, underfitting and overfitting with G = 2, G = 3, respectively. of 2,000 and using ECR for relabelling. Results are shown in [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: The error log(Zˆ) − log(Z) of the THAMES and bridge sampling for different values of T; on the left: the unconstrained Gaussian mixture model with G = 5, d = 6; on the right: the constrained Gaussian mixture model with G = 15, d = 5. Results The results are shown in […

**Figure 5.** Figure 5: Estimates from fully symmetric and random permutation bridge- (BSF, BSR), [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

**Figure 6.** Figure 6: Overlap graphs of the galaxies data set. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]

**Figure 7.** Figure 7: The values of µg, σg and W for a sample (µ (1) g , σ (1) g ), . . . ,(µ (T) g , σ (T) g ) from the posterior of the galaxy dataset of size T = 10000 and different values of G; the acceptance regions of QDA are shown via a grid in the background respectively, by permuti…

**Figure 8.** Figure 8: A toy example to illustrate the computation of the THAMES; the points that [PITH_FULL_IMAGE:figures/full_fig_p031_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

63 extracted references · 62 canonical work pages

[1]

Ameijeiras-Alonso, J., R. M. Crujeiras, and A. Rodríguez-Casal (2021). multimode: An R package for mode assessment.Journal of Statistical Software 97(9), 1–32. 13, 16

work page 2021
[2]

(2023).gor: Algorithms for the Subject Graphs and Network Optimiza- tion

Asensio, C. (2023).gor: Algorithms for the Subject Graphs and Network Optimiza- tion. R package version 1.0. 6

work page 2023
[3]

Bacharoglou, A. (2010). Approximation of probability distributions by convex mixtures of Gaussian measures. Proceedings of the American Mathematical Soci- ety 138(7), 2619–2628. 27

work page 2010
[4]

Baudry, J.-P., A. E. Raftery, G. Celeux, K. Lo, and R. Gottardo (2010). Com- bining mixture components for clustering. Journal of computational and graphical statistics 19(2), 332–353. 7, 15

work page 2010
[5]

Celeux, A

Bensmail, H., G. Celeux, A. E. Raftery, and C. P. Robert (1997). Inference in model-based cluster analysis.statistics and Computing 7, 1–10. 12, 39

work page 1997
[6]

van Mechelen, and A

Berkhof, J., I. van Mechelen, and A. Gelman (2003). A Bayesian approach to the selection and testing of mixture models.Statistica Sinica 13(2), 423–442. 2, 5, 16

work page 2003
[7]

Celeux, T

Bouveyron, C., G. Celeux, T. B. Murphy, and A. E. Raftery (2019).Model-based clustering and classification for data science: with applications in R, Volume 50. Cam- bridge University Press. 2

work page 2019
[8]

Gelman, M

Carpenter, B., A. Gelman, M. D. Hoffman, D. Lee, B. Goodrich, M. Betancourt, M. Brubaker, J. Guo, P. Li, and A. Riddell (2017). Stan: A probabilistic programming language. Journal of Statistical Software 76(1), 1–32. 36

work page 2017

Show all 63 references

[9]

Frühwirth-Schnatter, and C

Celeux, G., S. Frühwirth-Schnatter, and C. P. Robert (2019). Model selection for mixture models–perspectives and strategies. In Handbook of mixture analysis, pp. 117–154. Chapman and Hall/CRC. 1, 2, 13, 39, 40

2019
[10]

Hurn, and C

Celeux, G., M. Hurn, and C. P. Robert (2000). Computational and inferential difficulties with mixture posterior distributions.Journal of the American Statistical Association 95(451), 957–970. 8

2000
[11]

Chib, S. (1995). Marginal likelihood from the Gibbs output.Journal of the Amer- ican Statistical Association 90(432), 1313–1321. 2

1995
[12]

Coraggio, L. and P. Coretto (2023). Selecting the number of clusters, clustering models, and algorithms. a unifying approach based on the quadratic discriminant score. Journal of Multivariate Analysis 196, 105181. 15

2023
[13]

Fraley, C. and A. E. Raftery (2007). Bayesian regularization for normal mixture estimation and model-based clustering.Journal of classification 24(2), 155–181. 39

2007
[14]

Frühwirth-Schnatter, S. (2001). Markov chain Monte Carlo estimation of classical and dynamic switching and mixture models. Journal of the American Statistical Association 96(453), 194–209. 4 18 Marginal Likelihoods for Multivariate Mixture Models via THAMES

2001
[15]

Frühwirth-Schnatter, S. (2004). Estimating marginal likelihoods for mixture and Markov switching models using bridge sampling techniques.Econometrics Journal 7, 143–167. 2

2004
[16]

Frühwirth-Schnatter, S. (2006). Finite mixture and Markov switching models. Springer. 1, 28, 36, 38, 45

2006
[17]

Keepingthebalance—bridgesamplingformarginal likelihood estimation in finite mixture, mixture of experts and Markov mixture mod- els

Frühwirth-Schnatter,S.(2019). Keepingthebalance—bridgesamplingformarginal likelihood estimation in finite mixture, mixture of experts and Markov mixture mod- els. Brazilian Journal of Probability and Statistics 33(4), 706 – 733. 2

2019
[18]

Bayesianmodelchoice:asymptoticsandexact calculations

Gelfand,A.E.andD.K.Dey(1994). Bayesianmodelchoice:asymptoticsandexact calculations. Journal of the Royal Statistical Society: Series B (Methodological) 56, 501–514. 26

1994
[19]

Gelman, A., J. B. Carlin, H. S. Stern, and D. B. Rubin (1995).Bayesian data analysis. Chapman and Hall/CRC. 45

1995
[20]

Geweke, J. (2007). Interpretation and inference in mixture models: Simple MCMC works. Computational Statistics & Data Analysis 51(7), 3529–3550. 8

2007
[21]

Ghojogh, B. and M. Crowley (2019). Linear and quadratic discriminant analysis: Tutorial. 8

2019
[22]

Ghosal, S. (2000). Asymptotic normality of posterior distributions for exponential families when the number of parameters tends to infinity.Journal of Multivariate Analysis 74, 49–68. 8

2000
[23]

Goldfarb, D. and A. Idnani (1983). A numerically stable dual method for solving strictly convex quadratic programs.Mathematical programming 27(1), 1–33. 35

1983
[24]

Goldfarb, D. and A. Idnani (2006). Dual and primal-dual methods for solving strictly convex quadratic programs. InNumerical Analysis: Proceedings of the Third IIMAS Workshop Held at Cocoyoc, Mexico, January 1981, pp. 226–239. Springer. 35

2006
[25]

Gronau, Q. F., H. Singmann, and E.-J. Wagenmakers (2020). bridgesampling: An R package for estimating normalizing constants.Journal of Statistical Software 92(10), 1–29. 2, 11

2020
[26]

Gruen, B. (2023). bayesmix: Bayesian Mixture Models with JAGS. R package version 0.7-6. 40

2023
[27]

Grün, B. and F. Leisch (2009). Dealing with label switching in mixture models under genuine multimodality.Journal of Multivariate Analysis 100(5), 851–861. 27

2009
[28]

Hairault, A., C. P. Robert, and J. Rousseau (2022). Evidence estimation in finite and infinite mixture models and applications. 2, 45

2022
[29]

Hastings, W. K. (1970). Monte Carlo sampling methods using Markov chains and their applications. Biometrika 57(1), 97–109. 2

1970
[30]

Heyde, C. C. and I. M. Johnstone (1979). On asymptotic posterior normality for stochastic processes. Journal of the Royal Statistical Society: Series B (Methodologi- cal) 41, 184–189. 8 M. Metodiev et al. 19

1979
[31]

Theory of Probability(3rded.)

Jeffreys,H.(1961). Theory of Probability(3rded.). Oxford,U.K.:OxfordUniversity Press. 1

1961
[32]

(2013).Probability theory: a comprehensive course

Klenke, A. (2013).Probability theory: a comprehensive course. Springer Science & Business Media. 25

2013
[33]

Knuth, D. E. and J. L. Szwarcfiter (1974). A structured program to generate all topological sorting arrangements.Information Processing Letters 2(6), 153–157. 35

1974
[34]

Lee, J. E. and C. P. Robert (2016). Importance sampling schemes for evidence approximation in mixture models.Bayesian Analysis 11(2), 573 – 597. 2, 16

2016
[35]

Lenk, P. J. and W. S. DeSarbo (2000). Bayesian inference for finite mixtures of generalized linear models with random effects.Psychometrika 65(1), 93–119. 8

2000
[36]

Liver Disorders

Liver-Disorders (2016). Liver Disorders. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C54G67. 15

2016 doi
[37]

Martino, D

Llorente, F., L. Martino, D. Delgado, and J. Lopez-Santiago (2023). Marginal like- lihood computation for model selection and hypothesis testing: An extensive review. SIAM Review 65, 3–58. 1

2023
[38]

Meng, X.-L. and W. H. Wong (1996). Simulating ratios of normalizing constants via a simple identity: A theoretical exploration.Statistica Sinica 6(4), 831–860. 2, 11

1996
[39]

Metodiev, M., N. J. Irons, and M. Perrot-Dockès (2025).thamesmix: Truncated Harmonic Mean Estimator of the Marginal Likelihood for Mixtures. R package version 0.1.1. 16

2025
[40]

easily computed marginal likelihoods for multivariate mixture models using the thames estimator

Metodiev,M.,N.J.Irons,M.Perrot-Dockès,P.Latouche,andA.E.Raftery(2025). Supplement to "easily computed marginal likelihoods for multivariate mixture models using the thames estimator". 3, 5, 6, 9, 10, 11, 15

2025
[41]

Perrot-Dockès, S

Metodiev, M., M. Perrot-Dockès, S. Ouadah, N. J. Irons, P. Latouche, and A. E. Raftery (2024). Easily computed marginal likelihoods from posterior simulation using the THAMES estimator.Bayesian Analysis (advance publication), 1 – 28. 1, 2, 3, 5, 6, 27, 28, 29, 33, 34, 35

2024
[42]

Metropolis, N., A. W. Rosenbluth, M. N. Rosenbluth, A. H. Teller, and E. Teller (1953). Equation of state calculations by fast computing machines.Journal of Chem- ical Physics 21, 1087–1092. 2

1953
[43]

Miller, J. W. (2021). Asymptotic normality, concentration, and coverage of gener- alized posteriors. The Journal of Machine Learning Research 22, 7598–7650. 8

2021
[44]

Neal, R. M. (1996). Sampling from multimodal distributions using tempered tran- sitions. Statistics and computing 6, 353–366. 2

1996
[45]

marginal likelihood from the Gibbs output

Neal, R. M. (1999). Erroneous results in “marginal likelihood from the Gibbs output”. Unpublished manuscript. 2, 13

1999
[46]

Nobile, A. (2004). On the posterior distribution of the number of components in a finite mixture. The Annals of Statistics 32(5), 2044 – 2073. 3, 10, 32 20 Marginal Likelihoods for Multivariate Mixture Models via THAMES

2004
[47]

Nobile, A. (2007). Bayesian finite mixtures: a note on prior specification and pos- terior computation. arXiv preprint arXiv:0711.0458. 2, 3, 10, 32

2007 arXiv
[48]

Nowicki, K. and T. A. B. Snijders (2001). Estimation and prediction for stochastic blockstructures. Journal of the American statistical association 96(455), 1077–1087. 16

2001
[49]

Papastamoulis, P. and G. Iliopoulos (2010). An artificial allocations based solution to the label switching problem in Bayesian analysis of mixtures of distributions. Journal of Computational and Graphical Statistics 19(2), 313–331. 2, 5

2010
[50]

Ntzoufras, and E

Perrakis, K., I. Ntzoufras, and E. G. Tsionas (2014). On the use of marginal posteriors in marginal likelihood estimation via importance sampling.Computational Statistics & Data Analysis 77, 54–69. 2

2014
[51]

Reichl, J. (2020). Estimating marginal likelihoods from the posterior draws through a geometric identity.Monte Carlo Methods and Applications 26(3), 205–221. 2, 3, 4, 5, 11, 12, 34, 37

2020
[52]

Richardson, S. and P. J. Green (1997). On Bayesian analysis of mixtures with an unknown number of components (with discussion).Journal of the Royal Statistical Society Series B: Statistical Methodology 59(4), 731–792. 8, 13, 39

1997
[53]

Robert, C. P., D. Wraith, P. M. Goggans, and C.-Y. Chan (2009). Computational methods for Bayesian model choice. In AIP Conference Proceedings, pp. 251–262. AIP. 3

2009
[54]

Rufo, M. J., J. Martín, and C. Pérez (2010). New approaches to compute Bayes factor in finite mixture models. Computational statistics & data analysis 54(12), 3324–3335. 2

2010
[55]

Fraley, T

Scrucca, L., C. Fraley, T. B. Murphy, and A. E. Raftery (2023). Model-Based Clustering, Classification, and Density Estimation Using mclust in R. Chapman and Hall/CRC. 16

2023
[56]

Shen, X. (2002). Asymptotic normality of semiparametric and nonparametric pos- terior distributions. Journal of the American Statistical Association 97, 222–235. 8

2002
[57]

Sims, C. A., D. F. Waggoner, and T. Zha (2008). Methods for inference in large multiple-equation Markov-switching models.Journal of Econometrics 146(2), 255–

2008
[58]

RStan: the R interface to Stan

Stan Development Team (2022). RStan: the R interface to Stan. R package version 2.21.7. 2

2022
[59]

(1997).Bayesian methods for mixtures of normal distributions

Stephens, M. (1997).Bayesian methods for mixtures of normal distributions. Phd thesis, University of Oxford. Available at https://stephenslab.uchicago.edu/ assets/papers/DPhilMS.pdf. 8

1997
[60]

Stephens, M. (2000). Dealing with label switching in mixture models.Journal of the Royal Statistical Society: Series B (Statistical Methodology) 62(4), 795–809. 2, 5 M. Metodiev et al. 21

2000
[61]

Turlach, B. A. and A. Weingessel (2019).quadprog: Functions to Solve Quadratic Programming Problems. R package version 1.5-8. 35

2019
[62]

Wang, Y. J. and G. Y. Wong (1987). Stochastic blockmodels for directed graphs. Journal of the American Statistical Association 82(397), 8–19. 16

1987
[63]

acidity”, “enzyme

Wilcox, R. R. (2011). Introduction to robust estimation and hypothesis testing. Academic press. 34 22 Marginal Likelihoods for Multivariate Mixture Models via THAMES Supplementary Material Table of contents Supplement A: proofs 23 A.1 Proving that an optimal truncation set is ...

2011

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Ameijeiras-Alonso, J., R. M. Crujeiras, and A. Rodríguez-Casal (2021). multimode: An R package for mode assessment.Journal of Statistical Software 97(9), 1–32. 13, 16

work page 2021

[2] [2]

(2023).gor: Algorithms for the Subject Graphs and Network Optimiza- tion

Asensio, C. (2023).gor: Algorithms for the Subject Graphs and Network Optimiza- tion. R package version 1.0. 6

work page 2023

[3] [3]

Bacharoglou, A. (2010). Approximation of probability distributions by convex mixtures of Gaussian measures. Proceedings of the American Mathematical Soci- ety 138(7), 2619–2628. 27

work page 2010

[4] [4]

Baudry, J.-P., A. E. Raftery, G. Celeux, K. Lo, and R. Gottardo (2010). Com- bining mixture components for clustering. Journal of computational and graphical statistics 19(2), 332–353. 7, 15

work page 2010

[5] [5]

Celeux, A

Bensmail, H., G. Celeux, A. E. Raftery, and C. P. Robert (1997). Inference in model-based cluster analysis.statistics and Computing 7, 1–10. 12, 39

work page 1997

[6] [6]

van Mechelen, and A

Berkhof, J., I. van Mechelen, and A. Gelman (2003). A Bayesian approach to the selection and testing of mixture models.Statistica Sinica 13(2), 423–442. 2, 5, 16

work page 2003

[7] [7]

Celeux, T

Bouveyron, C., G. Celeux, T. B. Murphy, and A. E. Raftery (2019).Model-based clustering and classification for data science: with applications in R, Volume 50. Cam- bridge University Press. 2

work page 2019

[8] [8]

Gelman, M

Carpenter, B., A. Gelman, M. D. Hoffman, D. Lee, B. Goodrich, M. Betancourt, M. Brubaker, J. Guo, P. Li, and A. Riddell (2017). Stan: A probabilistic programming language. Journal of Statistical Software 76(1), 1–32. 36

work page 2017