betaFeedback

WorksMathematics

Likelihood-Free Schrödinger–Föllmer Sampling for Bayesian Inference with Fixed-Proposal Reuse

Abstract

We study Bayesian posterior sampling when the likelihood is an expectation that is costly to evaluate. We construct a likelihood-free Schrödinger–Föllmer sampler whose drift is written as a ratio of expectations over Gaussian variables and simulator outputs and is estimated directly from joint draws of the two. Importance sampling with a fixed proposal allows the simulator outputs to be generated once and reused across all time steps and trajectories. For the likelihood-free Schrödinger–Föllmer sampler, which draws new Monte Carlo samples at each step, we bound the terminal Wasserstein error under regularity and moment assumptions, separating time-discretization and integration errors. For the likelihood-free Schrödinger–Föllmer sampler with sample reuse, we prove that the output converges to the posterior in the Wasserstein distance, assuming only a finite posterior second moment. Numerical experiments illustrate the reduction in simulator calls and the benefits of randomized quasi-Monte Carlo for drift estimation.

The work of Z. He was supported by the National Natural Science Foundation of China under Grant 12671500 and by the Guangdong Basic and Applied Basic Research Foundation under Grant 2025A1515011888.

Introduction

We consider Bayesian inverse problems, in which unknown model parameters are inferred from observed data. The Bayesian approach combines prior information with a likelihood model and quantifies uncertainty through the posterior distribution [1]. This formulation is used, for example, for parameter estimation in differential equations [2] and in population genetics [3]. We focus on likelihoods given by conditional expectations that are costly to evaluate, and aim to reduce the simulation cost of posterior sampling.

Let x∈Rdx \in\mathbb{R}^{d} denote the unknown parameter. The posterior density is

π(x)=π0(x)L(x)Z,Z=∫Rdπ0(x)L(x) dx,(1)\pi(x) = \frac{\pi_{0}(x)L(x)}{\mathcal{Z}}, \qquad\mathcal{Z} = \int_{\mathbb{R}^{d}} \pi_{0}(x)L(x)\,\mathrm{d}x, \tag*{(1)}

where π0\pi_{0} is the prior density, LL is the likelihood or a specified likelihood approximation, and 0<Z<∞0 < \mathcal{Z} < \infty.

We consider targets for which

L(x)=Eξ∣x[ψ(x,ξ)],ξ∼p(⋅∣x),ψ(x,ξ)≥0.(2)L(x) = \mathbb{E}_{\xi\mid x}[\psi(x,\xi)], \qquad\xi\sim p(\cdot\mid x), \qquad\psi(x,\xi) \ge0. \tag*{(2)}

Here ξ\xi denotes simulator outputs or auxiliary variables, p(⋅∣x)p(\cdot\mid x) is their conditional distribution given xx, and ψ(x,ξ)\psi(x,\xi) is a computable nonnegative weight. We assume that ξ∼p(⋅∣x)\xi\sim p(\cdot\mid x) can be simulated, but not that the density p(ξ∣x)p(\xi\mid x) can be evaluated.

Two examples illustrate the representation in (I.2). In approximate Bayesian computation (ABC), simulated data receive larger weights when they agree more closely with the observations; averaging these weights gives an approximate likelihood [3, 4]. In latent-variable models, averaging the conditional likelihood over the unobserved variables gives the exact likelihood, which can therefore be estimated without bias by simulation [5]. Section II describes both constructions in detail.

Our goal is to generate approximately unweighted samples from π\pi when LL cannot be evaluated directly; such samples give any posterior summary, such as EX∼π[h(X)]\mathbb{E}_{X\sim\pi}[h(X)], without reweighting. Representation (I.2) also permits importance-sampling estimates of posterior expectations, but these are weighted averages rather than posterior samples. Equally weighted samples can be obtained by resampling the weighted points, but they are then confined to those points, and when the importance weights degenerate, which typically happens in high dimensions or for concentrated posteriors, only a few distinct points remain [6]. On the other hand, estimating (I.2) at every parameter value visited by a sampler can require many costly simulations, so we seek accurate posterior samples from fewer simulator evaluations.

Most methods for intractable likelihoods are based on Markov chain or sequential Monte Carlo. ABC compares simulated and observed data through a kernel and is implemented by rejection sampling [3], Markov chain Monte Carlo (MCMC) [7], or sequential Monte Carlo [8]; see Marin et al. [9] and Sisson et al. [4] for reviews. Synthetic likelihood replaces the likelihood of the summaries by a Gaussian fitted to simulations [10], and pseudo-marginal MCMC replaces the likelihood by a nonnegative unbiased estimator while still targeting the exact posterior [5, 11]. These MCMC-based methods require new simulations at every proposed parameter and produce correlated output, and the noise in the likelihood estimate must be tuned to avoid sticky chains [12]. Neural simulation-based inference instead trains a conditional density estimator on simulations [13], at the cost of network design and training.

Diffusion-based samplers transport a simple reference distribution to the target in finite time, so independent trajectories can be run in parallel without burn-in [14, 15]. Most of them learn the drift with neural networks. We build on Schrödinger–Föllmer sampling (SFS) [16, 17], a diffusion-based sampler whose drift is an explicit Gaussian expectation and can therefore be estimated by Monte Carlo without training.

SFS generates samples from a prescribed target by simulating the diffusion

dXt=b(Xt,t) dt+dBt,X0=0,(3)\mathrm{d}X_t = b(X_t,t)\,\mathrm{d}t + \mathrm{d}B_t,\qquad X_0=0, \tag*{(3)}

where BtB_t is a standard dd-dimensional Brownian motion. Let φ\varphi be the density of B1∼N(0,Id)B_1 \sim\mathcal{N}(0,I_d), which serves as a Gaussian reference, set f=π/φf=\pi/\varphi, and choose

b(x,t)=EZ[f(Z)(Z−x)](1−t)EZ[f(Z)],Z∼N(x,(1−t)Id),0≤t<1.(4)b(x,t)=\frac{\mathbb{E}_Z[f(Z)(Z-x)]}{(1-t)\mathbb{E}_Z[f(Z)]},\qquad Z\sim\mathcal{N}(x,(1-t)I_d),\qquad0\leq t<1. \tag*{(4)}

Under suitable regularity and integrability conditions, this choice gives X1∼πX_1\sim\pi [16, 17]; the resulting process solves the Schrödinger bridge problem between the point mass at the origin and π\pi [16, 18]. Since f(Z)=π0(Z)L(Z)/(Zφ(Z))f(Z)=\pi_0(Z)L(Z)/(\mathcal{Z}\varphi(Z)), the unknown normalizing constant Z\mathcal{Z} cancels in the ratio. However, ff still involves LL, which cannot be evaluated.

We combine (I.2) with iterated expectation to write the drift as a ratio of joint expectations over a Gaussian variable ZZ and an auxiliary variable ξ∼p(⋅∣Z)\xi\sim p(\cdot\mid Z). The drift can then be estimated directly from simulated pairs (Z,ξ)(Z,\xi). § 3 gives the derivation and the two resulting samplers.

Several works study the construction and numerical approximation of SFS. Huang et al. [16] introduced the sampler and analyzed its time discretization. Jiao et al. [17] established quadratic Wasserstein error bounds under smoothness and boundedness assumptions on the target density ratio, separating time-discretization and Monte Carlo (MC) errors. The MC estimator analyzed by Jiao et al. [17] writes the drift as EZ[∇f(Z)]/EZ[f(Z)]\mathbb{E}_Z[\nabla f(Z)]/\mathbb{E}_Z[f(Z)] and thus requires the gradient of the density ratio. In (I.4), the derivative is instead moved onto the Gaussian kernel, so only values of ff are needed; this is what later allows ff to be replaced by simulator outputs. Their MC error bound therefore does not cover our estimator.

Drift approximation is also related to importance sampling with random weights. Self-normalized importance sampling estimates a ratio of integrals using a common set of samples; its consistency and asymptotic properties are discussed in Owen [19 Section 9.2]. Llorente et al. [20] studied noisy importance sampling and the effect of random target evaluations on proposal design. Riel and Bischoff [21] used importance sampling and particle ensembles to approximate gradients of log densities without target gradients. Importance sampling with mixture proposals was studied by Hesterberg [22] and Owen and Zhou [23].

Quasi-Monte Carlo (QMC) and randomized QMC (RQMC) integration rules can improve on the MC error under suitable regularity assumptions [24]–[26]. For the SFS drift, however, the integrands depend on the current state and time, and reusing the same samples makes the subsequent states depend on them. An integration bound at a fixed state therefore does not by itself yield a bound on the terminal distribution.

The works above provide diffusion error estimates, importance sampling constructions, and integration rules, but they do not give a sampler for (I.2) that avoids estimating the likelihood at every parameter value. Our contributions are as follows.

  • Likelihood-free drift. Using iterated expectation, we write the SFS drift as a ratio of joint expectations over Gaussian variables and simulator outputs (III.11). The drift can thus be estimated from simulated pairs by MC or RQMC, without likelihood values or simulator derivatives.

  • Likelihood-free SFS. Estimating this drift with new samples at each time step gives likelihood-free SFS. Theorem IV.1 bounds its terminal quadratic Wasserstein distance W2W_2 error by O(h+Λh/m)O\left(\sqrt{h}+\sqrt{\Lambda_h/m}\right), where hh is the time step, mm is the number of samples per step, and Λh\Lambda_h, defined in (IV.3), accumulates the conditional mean-square drift-estimation error over the time grid.

  • Likelihood-free SFS with sample reuse. Importance sampling with a fixed proposal lets one set of MM simulated pairs serve all time steps and trajectories, reducing the number of simulator calls from NKmNKm to MM; in the Gaussian benchmark, 8192 calls replace about 10710^7. Theorem IV.2 proves that the output converges to the posterior in the quadratic Wasserstein distance W2W_2 under only a finite posterior second moment, and Theorem IV.3 gives a central limit theorem for the drift estimator that guides the choice of proposal.

This paper is organized as follows. Section II describes the problem setting and the likelihood representations. Section III introduces the two samplers. Section IV gives terminal error and convergence results for both, followed by the pointwise drift limit and the proposal-variance analysis. Section V reports numerical experiments, and Section VI concludes. Appendices A and B contain the proofs and supporting calculations.

Problem setting

Let x∈Rdx \in\mathbb{R}^{d} denote the unknown parameter and let yy denote the observed data. Since yy is fixed throughout, we suppress the dependence on yy in the notation for the likelihood and the posterior. We consider the posterior density

π(x)=π0(x)L(x)Z,Z=∫Rdπ0(u)L(u) du,(5)\pi(x) = \frac{\pi_{0}(x)L(x)}{\mathcal{Z}}, \qquad\mathcal{Z} = \int_{\mathbb{R}^{d}} \pi_{0}(u)L(u)\,\mathrm{d}u, \tag*{(5)}

where π0\pi_{0} is a prior density and 0<Z<∞0 < \mathcal{Z} < \infty. Throughout this section, LL is a nonnegative function defining the target; it may be the exact likelihood or a specified approximation, such as an ABC likelihood.

We assume the representation

L(x)=Eξ∣x[ψ(x,ξ)]=∫Ξψ(x,ξ) p(dξ∣x),ψ(x,ξ)≥0.(6)L(x) = \mathbb{E}_{\xi\mid x}[\psi(x,\xi)] = \int_{\Xi} \psi(x,\xi)\,p(\mathrm{d}\xi\mid x), \qquad\psi(x,\xi) \ge0. \tag*{(6)}

Here Ξ\Xi is the space of simulator outputs or auxiliary variables, p(dξ∣x)p(\mathrm{d}\xi\mid x) is their conditional probability law, and ψ(x,ξ)\psi(x,\xi) is a measurable, computable weight; the expectation is assumed finite for all parameter values considered. When this law has a density with respect to a reference measure, we also write p(ξ∣x)p(\xi\mid x) for that density. The samplers require draws from this law but never evaluate its density.

In ABC [3], [27], the model is a simulator: for given xx, it generates synthetic data ξ∼p(⋅∣x)\xi\sim p(\mathord{\cdot} \mid x) in the same space Ξ\Xi as the observed data yy, but the likelihood p(y∣x)p(y \mid x) of the observed data cannot be evaluated. ABC therefore compares the simulated data ξ\xi with yy. Let S:Ξ→RrS : \Xi\to\mathbb{R}^{r} be a summary map [28], where rr is the number of summary statistics. The vectors S(ξ)S(\xi) and S(y)S(y) may contain, for example, sample means or variances. We set

ψ(x,ξ)=Kδ(S(ξ),S(y)),ξ∼p(⋅∣x),(7)\psi(x,\xi) = K_{\delta}(S(\xi), S(y)), \qquad\xi\sim p(\mathord{\cdot} \mid x), \tag*{(7)}

where the nonnegative kernel KδK_{\delta} gives larger weights to closer matches, and its bandwidth δ>0\delta> 0 controls how much mismatch is tolerated. Equation (II.2) then defines the approximate likelihood LABC(x)=Eξ∣x[Kδ(S(ξ),S(y))]L_{\mathrm{ABC}}(x) = \mathbb{E}_{\xi\mid x}[K_{\delta}(S(\xi), S(y))]. The resulting ABC posterior, proportional to π0(x)LABC(x)\pi_{0}(x)L_{\mathrm{ABC}}(x), approximates the exact posterior, proportional to π0(x)p(y∣x)\pi_{0}(x)p(y \mid x), and has two sources of error. First, it conditions on the summaries S(y)S(y) instead of the full data yy, which loses information unless SS is sufficient. Second, the kernel also gives weight to simulated summaries that are only close to S(y)S(y); under mild conditions this error vanishes as δ→0\delta\to0.

A second example is a latent-variable model [5], in which unobserved random variables ξ\xi affect the data; for instance, repeated measurements on one individual may share an unobserved individual effect. If ξ\xi were observed, the likelihood of yy given (x,ξ)(x,\xi) would be computable; we take this conditional likelihood as ψ(x,ξ)\psi(x,\xi). Averaging it over the unobserved ξ∼p(⋅∣x)\xi\sim p(\mathord{\cdot} \mid x) gives the likelihood L(x)L(x) in (II.2), so here LL is the exact likelihood, with no ABC approximation. More generally, ξ\xi can be the random inputs of any nonnegative unbiased estimator of L(x)L(x).

By contrast, for a model y=G(x)+εy = G(x) + \varepsilon with a deterministic forward map GG and additive noise with density ρ\rho, independent of xx, the likelihood L(x)=ρ(y−G(x))L(x) = \rho(y - G(x)) can be evaluated directly whenever GG can, and no auxiliary simulation is needed.

A direct Monte Carlo approximation of (II.2) is

L^m(x)=1m∑i=1mψ(x,ξi),ξi∼i.i.d.p(⋅∣x).(8)\widehat{L}_m(x) = \frac{1}{m}\sum_{i=1}^{m}\psi(x,\xi_i), \qquad\xi_i \overset{\mathrm{i.i.d.}}{\sim} p(\cdot\mid x). \tag*{(8)}

For fixed xx, this estimator is unbiased, E[L^m(x)∣x]=L(x)\mathbb{E}[\widehat{L}_m(x)\mid x] = L(x). Its drawback is cost. Each likelihood value requires mm simulations, and a posterior sampler needs the likelihood at many parameter values, for example at every proposed state of a Markov chain. The total number of simulations is therefore mm times the number of likelihood evaluations, and mm must be large enough to control the noise in each estimate.

Representation (II.2) also allows posterior expectations to be estimated without a sampler. Given a proposal density qq, draw Zi∼qZ_i \sim q and ξi∼p(⋅∣Zi)\xi_i \sim p(\cdot\mid Z_i), i=1,…,Mi = 1,\ldots,M, and set vi=π0(Zi)ψ(Zi,ξi)/q(Zi)v_i = \pi_0(Z_i)\psi(Z_i,\xi_i)/q(Z_i). Since Eq[vig(Zi)]=ZEπ[g(X)]\mathbb{E}_q[v_i g(Z_i)] = Z\mathbb{E}_{\pi}[g(X)], the self-normalized estimate ∑ivig(Zi)/∑ivi\sum_i v_i g(Z_i)/\sum_i v_i converges to Eπ[g(X)]\mathbb{E}_{\pi}[g(X)] [19 Chapter 9]. This yields weighted estimates of posterior expectations, not posterior samples. Resampling from the weighted empirical distribution ∑i(vi/∑jvj)δZi\sum_i(v_i/\sum_j v_j)\delta_{Z_i} gives equally weighted samples, but only at the points Z1,…,ZMZ_1,\ldots,Z_M, and the number of distinct points is governed by the effective sample size (∑ivi)2/∑ivi2(\sum_i v_i)^2/\sum_i v_i^2, which typically collapses as the dimension grows or as the posterior concentrates relative to qq [6].

The sampler in § 3 instead outputs unweighted samples and requires LL only through averages over parameter values. By (II.2), such averages are joint expectations over the parameter and the simulator output, and the simulations can be generated once by importance sampling and reused throughout the sampler.

Likelihood-free SFS methods

Section III-A recalls SFS and its Euler–Maruyama discretization, in which the only quantity that cannot be computed directly is the drift. Section III-B writes the drift as a ratio of joint expectations over Gaussian variables and simulator outputs. Sections III-C and III-D give two estimators of this ratio: likelihood-free SFS draws new samples at each step, and likelihood-free SFS with sample reuse uses importance sampling so that the samples are generated once and reused across all time steps and trajectories.

Schrödinger–Föllmer sampling and its time discretization

Let

References

  1. [1]A. M. Stuart, “Inverse problems: A Bayesian perspective,” Acta Numerica, vol. 19, pp. 451–559, 2010.DOI
  2. [2]M. Dashti and A. M. Stuart, “The Bayesian approach to inverse problems,” in Handbook of Uncertainty Quantification. Springer, 2017, pp. 311–428.
  3. [3]M. A. Beaumont, W. Zhang, and D. J. Balding, “Approximate Bayesian computation in population genetics,” Genetics, vol. 162, no. 4, pp. 2025–2035, 2002.pubmed.ncbi.nlm.nih.gov/12524368
  4. [4]S. A. Sisson, Y. Fan, and M. Beaumont, Eds., Handbook of Approximate Bayesian Computation. Chapman and Hall/CRC, 2018.
  5. [5]C. Andrieu and G. O. Roberts, “The pseudo-marginal approach for efficient Monte Carlo computations,” The Annals of Statistics, vol. 37, no. 2, pp. 697–725, 2009.DOI
  6. [6]S. Agapiou, O. Papaspiliopoulos, D. Sanz-Alonso, and A. M. Stuart, “Importance sampling: Intrinsic dimension and computational cost,” Statistical Science, vol. 32, no. 3, pp. 405–431, 2017.arxiv.org/abs/1511.06196
  7. [7]P. Marjoram, J. Molitor, V. Plagnol, and S. Tavare, “Markov chain Monte Carlo without likelihoods,” Proceedings of the National Academy of Sciences, vol. 100, no. 26, pp. 15 324–15 328, 2003.pubmed.ncbi.nlm.nih.gov/14663152
  8. [8]P. Del Moral, A. Doucet, and A. Jasra, “An adaptive sequential Monte Carlo method for approximate Bayesian computation,” Statistics and Computing, vol. 22, no. 5, pp. 1009–1020, 2012.DOI
  9. [9]J.-M. Marin, P. Pudlo, C. P. Robert, and R. J. Ryder, “Approximate Bayesian computational methods,” Statistics and Computing, vol. 22, no. 6, pp. 1167–1180, 2012.arxiv.org/abs/1101.0955
  10. [10]L. F. Price, C. C. Drovandi, A. Lee, and D. J. Nott, “Bayesian synthetic likelihood,” Journal of Computational and Graphical Statistics, vol. 27, no. 1, pp. 1–11, 2018.DOI
  11. [11]C. Andrieu, A. Doucet, and R. Holenstein, “Particle Markov chain Monte Carlo methods,” Journal of the Royal Statistical Society: Series B, vol. 72, no. 3, pp. 269–342, 2010.DOI
  12. [12]A. Doucet, M. K. Pitt, G. Deligiannidis, and R. Kohn, “Efficient implementation of Markov chain Monte Carlo when using an unbiased likelihood estimator,” Biometrika, vol. 102, no. 2, pp. 295–313, 2015.arxiv.org/abs/1210.1871
  13. [13]K. Cranmer, J. Brehmer, and G. Louppe, “The frontier of simulation-based inference,” Proceedings of the National Academy of Sciences, vol. 117, no. 48, pp. 30 055–30 062, 2020.DOI
  14. [14]Q. Zhang and Y. Chen, “Path integral sampler: A stochastic control approach for sampling,” in International Conference on Learning Representations, 2022.arxiv.org/abs/2111.15141
  15. [15]F. Vargas, W. Grathwohl, and A. Doucet, “Denoising diffusion samplers,” in International Conference on Learning Representations, 2023.arxiv.org/abs/2302.13834
  16. [16]J. Huang, Y. Jiao, L. Kang, X. Liao, J. Liu, and Y. Liu, “Schrödinger–Föllmer sampler,” IEEE Transactions on Information Theory, vol. 71, no. 2, pp. 1283–1299, 2025.DOI
  17. [17]Y. Jiao, L. Kang, Y. Liu, and Y. Zhou, “Convergence analysis of Schrödinger–Föllmer samplers without convexity,” Science China Mathematics, 2026.DOI
  18. [18]C. Léonard, “A survey of the Schrödinger problem and some of its connections with optimal transport,” Discrete and Continuous Dynamical Systems, vol. 34, no. 4, pp. 1533–1574, 2014.DOI
  19. [19]A. B. Owen, “Monte Carlo theory, methods and examples,” 2013, chapter 9: Importance sampling, Section 9.2. [Online]. Available: https://artowen.su.domains/mc/Ch-var-is.pdf
  20. [20]F. Llorente, L. Martino, J. Read, and D. Delgado-Gómez, “Optimality in noisy importance sampling,” Signal Processing, vol. 194, p. 108455, 2022.DOI
  21. [21]B. Riel and T. Bischoff, “Gradient-free score-based sampling methods with ensembles,” Applied Mathematical Modelling, vol. 147, p. 116224, 2025.DOI
  22. [22]T. Hesterberg, “Weighted average importance sampling and defensive mixture distributions,” Technometrics, vol. 37, no. 2, pp. 185–194, 1995.DOI
  23. [23]A. Owen and Y. Zhou, “Safe and effective importance sampling,” Journal of the American Statistical Association, vol. 95, no. 449, pp. 135–143, 2000.DOI
  24. [24]H. Niederreiter, Random Number Generation and Quasi-Monte Carlo Methods, ser. CBMS-NSF Regional Conference Series in Applied Mathematics. SIAM, 1992, vol. 63.
  25. [26]———, “Scrambled net variance for integrals of smooth functions,” The Annals of Statistics, vol. 25, no. 4, pp. 1541–1562, 1997.DOI
  26. [27]M. G. B. Blum, “Approximate Bayesian computation: A nonparametric perspective,” Journal of the American Statistical Association, vol. 105, no. 491, pp. 1178–1187, 2010.arxiv.org/abs/0904.0635
  27. [28]D. Prangle, “Summary statistics,” in Handbook of Approximate Bayesian Computation, S. A. Sisson, Y. Fan, and M. Beaumont, Eds. Chapman and Hall/CRC, 2018, pp. 125–152.
  28. [29]I. Karatzas and S. E. Shreve, Brownian Motion and Stochastic Calculus, 2nd ed., ser. Graduate Texts in Mathematics. New York: Springer, 1991, vol. 113.
  29. [30]A. B. Owen, “Halton sequences avoid the origin,” SIAM Review, vol. 48, no. 3, pp. 487–503, 2006.DOI
  30. [31]D. Ouyang, X. Wang, and Z. He, “Achieving high convergence rates by quasi-Monte Carlo and importance sampling for unbounded integrands,” SIAM Journal on Numerical Analysis, vol. 62, no. 5, pp. 2393–2414, 2024.DOI
  31. [32]S. Barber, J. Voss, and M. Webster, “The rate of convergence for approximate Bayesian computation,” Electronic Journal of Statistics, vol. 9, no. 1, pp. 80–105, 2015.arxiv.org/abs/1311.2038
  32. [33]D. T. Frazier, G. M. Martin, C. P. Robert, and J. Rousseau, “Asymptotic properties of approximate Bayesian computation,” Biometrika, vol. 105, no. 3, pp. 593–607, 2018.
  33. [34]M. Gelbrich, “On a formula for the L² Wasserstein metric between measures on Euclidean and Hilbert spaces,” Mathematische Nachrichten, vol. 147, no. 1, pp. 185–203, 1990.DOI
  34. [35]S. L. Cotter, G. O. Roberts, A. M. Stuart, and D. White, “MCMC methods for functions: Modifying old algorithms to make them faster,” Statistical Science, vol. 28, no. 3, pp. 424–446, 2013.DOI
  35. [36]M. G. B. Blum and O. François, “Non-linear regression models for approximate Bayesian computation,” Statistics and Computing, vol. 20, no. 1, pp. 63–73, 2010.DOI
  36. [37]N. Bonneel, J. Rabin, G. Peyré, and H. Pfister, “Sliced and Radon Wasserstein barycenters of measures,” Journal of Mathematical Imaging and Vision, vol. 51, no. 1, pp. 22–45, 2015.
  37. [38]U. Halekoh, S. Højsggaard, and J. Yan, “The R package deepack for generalized estimating equations,” Journal of Statistical Software, vol. 15, no. 2, pp. 1–11, 2006.
  38. [39]G. M. Fitzmaurice and N. M. Laird, “A likelihood-based method for analysing longitudinal binary responses,” Biometrika, vol. 80, no. 1, pp. 141–151, 1993.DOI
  39. [40]L. Tierney and J. B. Kadane, “Accurate approximations for posterior moments and marginal densities,” Journal of the American Statistical Association, vol. 81, no. 393, pp. 82–86, 1986.DOI
  40. [41]G. H. Golub and J. H. Welsch, “Calculation of Gauss quadrature rules,” Mathematics of Computation, vol. 23, no. 106, pp. 221–230, 1969.DOI

Paper details

vlnc:W_utwvc6qt
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents