Pith. sign in

REVIEW 3 major objections 5 minor 1 cited by

BASIL: Fast broadband line-rich spectral-cube fitting and image visualization via Bayesian quadrature

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read A 2-million-hour mapping job done in 180 hours

desk verdict A serious methods contribution for fast spectral-cube fitting, where the full pipeline is only validated on smooth Gaussian maps and the more realistic benchmark tests only the map-reconstruction block, leaving the headline speed claim for complex sources open. read the letter →

arxiv 2506.23269 v1 pith:WAEOO6ID submitted 2025-06-29 astro-ph.GA

Yuxin Lin , Masaki Adachi , Silvia Spezzano , Gordian Edenhofer , Vincent Eberle , Michael A. Osborne , Paola Caselli This is my paper · ORCID
classification astro-ph.GA
keywords spectral-cubefittinghotcoresBayesianquadratureactivelearningGaussianprocessstochasticvariationalinferenceLTEradiativetransfermolecularparametermaps
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 argues that mapping molecules across wideband spectral datacubes, a task typically requiring millions of hours of pixel-by-pixel fitting, can instead be done by fitting only a few hundred to a thousand carefully chosen pixels and predicting the entire map with a Gaussian process. The pixel choices are driven by Bayesian quadrature, which selects the locations that most reduce map-wide uncertainty. On a synthetic hot-core cube with 117 molecules and 40,000 pixels, the claim is that this recovers accurate maps of all 468 molecular parameters in about 180 hours, with visually reliable results after about 20 hours, at a root mean square error comparable to brute-force fitting. The paper presents this as a practical route to systematic analysis of the large, line-rich surveys expected from future ALMA upgrades.

What carries the argument

The load-bearing mechanism is the reformulation of parallel active-learning pixel selection as a Bayesian quadrature problem: the GP's predictive variance is treated as an unnormalized target distribution, and kernel quadrature picks a batch of points that best represents that distribution, effectively maximizing the reduction in integrated uncertainty across the map. The argument also relies on a Kronecker-structured multi-output Gaussian process that shares information across the 468 parameter maps, and on stochastic variational inference with a mixture likelihood that treats channels contaminated by unidentified lines as outliers.

What would settle it

Run the full BASIL pipeline on a realistic spectral cube with known ground-truth maps that contain sharp chemical boundaries, and check whether the GP-reconstructed maps reproduce those edges: if deviations at the boundaries exceed the roughly ten percent mean square error reported on the Gaussian benchmark maps, the smoothness assumption is violated.

Watch

Extended reading notes

Core claim

The central claim is that molecular parameter maps can be estimated without exhaustively fitting every pixel. The pipeline works in two stages: stochastic variational inference estimates the four LTE parameters per molecule at individual positions, and a multi-output Gaussian process then extrapolates those estimates over the spatial domain, greedily querying the pixels with the highest predictive variance through the batch Bayesian active learning method BASQ. In the synthetic benchmark of 117 molecules, each with excitation temperature, column density, centroid velocity, and line width, roughly 1,000 queried locations suffice to produce maps whose errors are comparable to full pixel-by-pixel fitting, while the compute time drops from an estimated two million hours to about 180 hours. The paper attributes the speed to sublinear convergence of the GP prediction error, and emphasizes that the framework simultaneously fits all molecules under a one-component LTE model to handle line blending.

Load-bearing premise

The Gaussian process assumes the true parameter maps are smooth with a stationary correlation length, so sharp boundaries, sub-beam structure, or nonstationary emission in real sources would be smoothed away and the demonstrated speed-accuracy trade-off would degrade.

Editorial extensions

If this is right

  • Any wideband spectral cube of a line-rich source can be mapped at a fraction of the compute, with the same pixel-selection and GP recipe reusable for other per-pixel fitting problems.
  • Systematic analyses of large surveys become feasible, enabling unbiased statistics of chemical abundances and kinematics across many hot cores and hot corinos rather than case studies.
  • The iterative workflow provides a natural quick-look mode: about 20 hours yields visually reliable maps, and continued iterations refine them, matching the data volume of next-generation ALMA observations.
  • The dense, continuous maps produced by the GP remove the jagged, pixel-to-pixel discontinuities that plague independent per-pixel optimizations.
  • The estimated two-million-hour MCMC baseline makes the speed gain concrete: the reported 180 hours is roughly four orders of magnitude faster under the paper's assumptions.

Reading between the lines

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

  • A direct test of the full pipeline would apply it to an observed datacube, such as the PILS data toward IRAS16293B, and compare the GP-produced maps against published column densities; the paper only validates the SVI step against those data, not the GP and BASQ extrapolation.
  • The smoothness assumption limits the method on sources with sharp chemical boundaries or sub-beam structure; using nonstationary or spatially varying kernels is a natural next step the paper does not test.
  • The framework is not specific to LTE or molecular line fitting; it could accelerate any spectral-cube analysis where the underlying parameter fields are spatially smooth, such as line decomposition in Galactic plane surveys.
  • Because the benchmark maps were generated as two-component Gaussians, the reported error rates may be optimistic; realistic emission with spiral-arm-like structure (as in the appendix test) would be a fairer measure of the full pipeline.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents BASIL, a workflow that combines stochastic variational inference (SVI) for per-pixel LTE spectral fitting with Gaussian-process-based Bayesian active learning (BASQ) to estimate molecular parameter maps from line-rich spectral datacubes. The method is benchmarked on a synthetic 200x200-pixel cube with 117 molecules and 138,016 frequency channels, claiming 468 parameter maps in about 180 hours, visually reliable maps after about 20 hours, and an RMSE comparable to ground truth while being orders of magnitude faster than pixel-by-pixel MCMC. The SVI step is additionally cross-validated against published PILS column densities.

Significance. If the headline claims hold, BASIL would be a practically valuable tool for the large wideband surveys expected from the ALMA Wideband Sensitivity Upgrade. The paper's strengths include a public code repository with a minimal working example, a synthetic benchmark with known ground truth, the use of a realistic hydrodynamical simulation in Appendix B for the map-reconstruction step, and a comparison of SVI-derived column densities against PILS results. The central methodological idea—replacing exhaustive pixel-by-pixel fitting with active learning over a GP surrogate—is sensible and the separation of the SVI and BASQ blocks is clearly described. However, the validation as presented supports the conclusion only under favorable smoothness assumptions, and some of the abstract-level claims go beyond what the experiments directly demonstrate.

major comments (3)
  1. [§6.1, Appendix B] The full-pipeline benchmark in §6.1 uses synthetic parameter maps generated as two-component multivariate normals, which is the ideal case for an RBF-kernel GP. Appendix B tests the BASQ/GP block on the more realistic V892 Tau simulation maps, but it isolates that block by replacing actual SVI outputs with ground-truth maps plus i.i.d. noise; no full-pipeline wall-clock time or RMSE including SVI error is reported. The paper itself states in §6.1 that the universal approximation property does not imply sample efficiency at the n≈1000 scale used here. Because the headline speed-accuracy trade-off depends on reconstructing 468 maps from roughly 1000 SVI-fitted pixels, the authors should run the full BASIL pipeline on non-smooth or simulated hot-core maps—for example, by feeding the V892 Tau maps through the actual LTE+SVI step—and report the resulting RMSE and time. Without such a test, the central claim is only demonstrated for maps that are essentially GP-friendly.
  2. [§4.1, Eqs. (8)–(9), §6.2] Equation (8) defines the RMSE as sqrt(mean(rSVI + rGP)), but the true squared error between the ground-truth map and the GP prediction is (y* - \hat f)^2 = rSVI + rGP + 2(y* - \hat y)(\hat y - \hat f). The cross term is omitted unless the SVI error and the GP error are uncorrelated, which is not established. Equation (9) further assumes the SVI estimation error is zero-mean Gaussian, yet §6.2 documents severe degeneracies for HNCO, HN13CO, and CH3COOH, with larger scatter in the low-Tex regime, so the zero-mean and Gaussianity assumptions are not obviously satisfied. Since Eq. (9) and the predictive-variance proxy in Eq. (10) are used for the stopping criterion and for uncertainty estimates on real data, the paper needs to justify or correct this decomposition and to assess the impact of biased or non-Gaussian SVI errors on both the GP surrogate and the active-learning acquisition.
  3. [Abstract, §6.3] The headline claim of being 'orders of magnitude faster' than traditional pixel-by-pixel MCMC rests on an unmeasured estimate of roughly 2e6 hours for the MCMC baseline, and the 'comparable RMSE' is not directly benchmarked against an MCMC fit; the synthetic experiments compare GP predictions with ground truth, not with MCMC results. The only measured comparison in §6.3 is the factor of about 50 relative to exhaustive SVI fitting (40,000 versus 1,000 queried locations). The authors should either provide the basis for the MCMC runtime estimate—including per-pixel runtime, number of walkers, burn-in, thinning, and convergence criteria—or reframe the speed and accuracy claims in terms of the measured SVI exhaustive-fit baseline.
minor comments (5)
  1. [§6.3] The statement that 'MSE values fall within the model prediction standard deviations and are approximately 10% of the ground-truth values' is ambiguous; the figures suggest the MSE is about 10% of the parameter range rather than 10% of the local ground-truth value, and this should be stated precisely.
  2. [Figure 9 caption] The caption says the MSE color scale is capped at 0.5 times the maximum model standard deviation, which is a nonstandard display choice; please clarify how this affects the visual impression of the MSE maps and whether the same cap is applied to all molecules.
  3. [§6.1] The removal of molecules with fewer than three transitions is an ad hoc cut that affects the claimed molecule count and the difficulty of the fitting problem; the paper should justify why three transitions is sufficient to constrain Nmol and Tex and state how many of the 117 entries are affected.
  4. [§5.2] The mixture likelihood for handling unidentified lines introduces an outlier probability as an additional optimized parameter, but the chosen prior or allowed range for this probability is not specified; adding this information would improve reproducibility.
  5. [Appendix B] The RMSE comparison in Figure B.2 appears to come from a single run; multiple random seeds or confidence intervals would strengthen the claim that BASQ outperforms random nearest-neighbor interpolation by a stable margin.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the central accuracy and speed claims are benchmarked against external synthetic ground truth; the only self-citation (BASQ convergence theorem) is not load-bearing.

full rationale

The paper's derivation chain is self-contained with respect to its accuracy claims. The GP is trained on SVI-inferred parameters and evaluated against externally generated ground-truth maps in Section 6 (synthetic datacube), not against quantities the GP itself produces by construction; Eq. (8) separates SVI error from GP error, and the reported MSE is computed against ground truths. The Appendix B benchmark, while isolating the GP+BASQ block and simulating SVI outputs as ground truth plus i.i.d. noise, is an honest partial test rather than a disguised prediction: the noisy values are external inputs, and the RMSE is measured against independent simulation maps. The claim of sublinear/exponential convergence relies on a published theorem from the authors' prior work (Adachi et al. 2022, Theorem 1), cited in Section 5.3; this is a self-citation, but it is not load-bearing because the empirical speedup is demonstrated on the synthetic cube and the theorem is a parameter-free mathematical result with its own proof. The main caveats are validity risks rather than circularity. Section 6.1 generates the benchmark maps as two-component multivariate normals, which are well matched to an RBF-kernel GP, and it explicitly warns that RBF universal approximation 'does not imply that the RBF kernel always provides better sample efficiency for function approximation compared to other kernels in practice.' Appendix B tests only the map-reconstruction block by adding i.i.d. noise to ground-truth maps, so it does not capture spatially correlated SVI bias or full-pipeline wall-clock. These concerns weaken external generalizability to non-smooth real hot-core maps, but no claimed result reduces to its own inputs by construction, and no fitted parameter is renamed as a prediction.

Assumptions & free parameters 5 free parameters · 7 assumptions · 0 invented entities

The paper introduces no new physical entities such as particles or forces. The added objects are statistical: the outlier mixture component and the Gaussian process model. The main load-bearing assumptions are the smoothness of parameter maps and the unbiasedness of the SVI error model, both of which are favorable for the benchmark but not yet demonstrated on real data.

free parameters (5)
  • GP kernel hyperparameters (RBF lengthscale, output scale, noise variance) = not reported in text
    Fit to queried SVI data by marginal likelihood MLE in Algorithm 1 step 4. The reconstruction quality and the active learning query selection both depend on these values.
  • SVI variational parameters (multivariate normal guide mean and covariance per pixel) = not reported in text
    Optimized via ELBO for each queried spectrum. All downstream GP maps inherit any bias or underestimated uncertainty from these per-pixel fits.
  • Batch size n = 50
    Chosen by the authors for the benchmark. It controls the trade-off between parallel SVI throughput and the information gained per iteration.
  • Outlier probability in the mixture likelihood = not reported in text
    Introduced to absorb unidentified lines. If this parameter is poorly constrained, the fitted line intensities can be biased.
  • Synthetic ground-truth map parameters (peak centers and variances of two Gaussian components) = drawn randomly per molecule
    These define the benchmark rather than the method, but their smoothness directly affects how easily the GP can reconstruct the maps.
assumptions (7)
  • domain assumption Local thermodynamic equilibrium with a single excitation temperature holds for all 117 molecules in hot cores and hot corinos.
    Section 2; the LTE model with four parameters per molecule is the physical model whose maps BASIL produces. If LTE fails, the fitted maps do not describe the gas.
  • domain assumption Line optical depths from different molecules simply add, and radiative coupling between blended species is neglected.
    Section 2; the paper explicitly omits line trapping across molecules, which matters in heavily blended spectra.
  • domain assumption Continuum-subtracted data are described by Tbg = 2.73 K, Tslope = 0, and tau_dust = 0.
    Section 2; real observations may retain continuum residuals and frequency-dependent slopes that violate this assumption.
  • ad hoc to paper The SVI estimation error is zero-mean Gaussian, so the GP predictive variance is a valid proxy for the total RMSE.
    Section 4.1, Eq. 9; this assumption justifies the stopping criterion and the reported estimated RMSE, but the paper does not demonstrate zero-mean unbiased SVI errors.
  • domain assumption Parameter maps are smooth enough for GP interpolation, and the chosen RBF kernel has sufficient capacity.
    Section 6.1 and Appendix B; the main benchmark uses two-component Gaussian maps, directly matching this assumption, while the more realistic simulation maps are only used for the BASQ step in isolation.
  • domain assumption Molecular constants from CDMS, JPL, and LSD are accurate for all entries used.
    Section 2; all spectra are computed from these catalogs, and catalog errors propagate into the fitted parameters.
  • ad hoc to paper Molecules with fewer than three transitions in the frequency band can be removed from the fit.
    Section 6; the paper removes them for identifiability, but such species may still emit and contribute to blending in real data.

how reviews work

0 comments
Cite this review

Pith. "Pith review of BASIL: Fast broadband line-rich spectral-cube fitting and image visualization via Bayesian quadrature." pith.science (2026). https://pith.science/paper/WAEOO6ID

@misc{pith2026250623269,
  author       = {Pith},
  title        = {Pith review of: BASIL: Fast broadband line-rich spectral-cube fitting and image visualization via Bayesian quadrature},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WAEOO6ID}},
  note         = {Machine review of arXiv:2506.23269}
}
read the original abstract

Mapping the spatial distributions and abundances of complex organic molecules in hot cores and hot corinos is crucial for understanding the astrochemical pathways and the inheritance of prebiotic material by nascent planetary systems. However, the line-rich spectra from these sources pose significant challenges for robustly fitting molecular parameters due to severe line blending and unidentified lines. We present an efficient framework, Bayesian Active Spectral-cube Inference and Learning (BASIL), for estimating molecular parameter maps for hundreds of molecules based on the local thermodynamic equilibrium (LTE) model, applied to wideband spectral datacubes of line-rich sources. We adopted stochastic variational inference to infer molecular parameters from spectra at individual positions, balancing between fitting accuracy and computational speed. For obtaining parameter maps, instead of querying every location or pixel, we introduced an active learning framework based on Bayesian quadrature and its parallelization. Specifically, we assessed and selected the locations or pixels of spectrum that are most informative for estimating the entire set of parameter maps by training a Gaussian processes model. By greedily selecting locations with maximum information gain, we achieve sublinear convergence. We benchmarked BASIL on a large synthetic datacube and demonstrated that it produces accurate 468 molecular parameter maps from 117 molecules within ~180 hours, orders of magnitude faster than traditional pixel-by-pixel fitting using Markov chain Monte Carlo methods, with visually reliable results emerging in just ~20 hours. Additional training iterations provide progressively more accurate results. This quick visualization meets the demands of big data in modern astronomical surveys.

Figures

Figures reproduced from arXiv: 2506.23269 by the authors.

Figure 1
Figure 1. Flow chart of BASIL. Kullback-Leibler (KL) divergence. In practice, instead of min￾imizing KL divergence, one commonly maximizes the evidence lower bound (ELBO) (essentially equivalent to the negative KL divergence) with an additional constant term, which represents a lower constraint on the logarithm of the marginal probability of the observations. Specifically, the formula for ELBO follows such a form of expectati… view at source ↗
Figure 2
Figure 2. Examples of loss variation as a function of the number of it￾erations for 50 SVI batch fitting results, showing average losses and standard deviations (shaded area). exact distribution and computational complexity. It is also pos￾sible to approximate a multimodal distribution using techniques such as bijective transformation to employ more expressive dis￾tribution families such as normalizing flows (Dinh et al. 2014… view at source ↗
Figure 3
Figure 3. Cross-validation results of Tex, Nmol, Vc , and σv for each molecule at 50 initial randomly selected positions. 336.8 336.9 337.0 337.1 337.2 337.3 Freq. (GHz) 2000 1500 1000 500 0 500 Tm b (K) Syn. Obs.×0.2 CH3OH, vt=0,1 c-C3H2 Ethylene oxide HC(O)NH2, v=0 HC(O)NH2, v12=1 HC-13-(O)NH2 HC(O)N-15-H2 DC(O)NH2 cis-HC(O)NHD trans-HC(O)NHD C2H5OH,v=0 t-HC-13-OOH a-C-13-H3CH2OH a-CH3CH2OD a-a-CH2DCH2OH a-s-CH2DCH2OH CH3SH… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Example of a synthetic spectrum in the synthetic datacube, showing all molecules with significant emitting lines (peak intensity >250 K), as an illustration of line blending, offset by 50 K. The composite spectrum is scaled by 0.2 for better comparison. In reality, che…
Figure 5
Figure 5. Figure 5: Example of a synthetic spectrum from the synthetic datacube. Subplots are arranged in chunks of increasing frequency. The synthetic observed spectrum is indicated as a thick gray line, and the SVI inferred model spectrum as a dashed red line. The synthetic spectra for …
Figure 6
Figure 6. Figure 6: Examples of 2D distributions of the ground-truth parameters, Tex, Nmol, Vc , and σv generated by a two-component normal distribution. The parameter ranges are all normalized to follow N(0, 1). ties for more than 130 molecules have been reported toward IRAS16293B, some …
Figure 7
Figure 7. Figure 7: First-iteration results of multi-output GP model predictions and predicted standard deviations for the four parameters, Tex, Nmol, Vc , and σv, after SVI fitting to 50 initially randomly selected positions and model training. The white dots mark the locations where the…
Figure 8
Figure 8. Figure 8: Sum of the GP model’s predictive variances across all parameter maps, after the first, fifth, and fifteenth iteration of training. The white dots indicate the queried locations, while the orange pluses indicate the next-iteration locations proposed by the GP model. -10…
Figure 9
Figure 9. Figure 9 [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: Cross-validation of the SVI-fitted column densities with the PILS results from the literature. The red square indicates the location of CH3 18OH, from which the column density of CH3OH is derived using a fixed 18O/ 16O ratio. The solid line indicates the equal line, a…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. End-to-end differentiable retrieval of molecular spectra using hydrodynamics, chemistry, and radiative transfer

    astro-ph.IM 2026-07 conditional novelty 6.0 of 10

    An end-to-end differentiable JAX pipeline couples 1D hydrodynamics, time-dependent chemistry, and radiative transfer, and recovers shock and rate parameters from synthetic HCO+ spectra.

Reference graph

Works this paper leans on

98 extracted references · 46 canonical work pages · cited by 1 Pith paper

  1. [1]

    2023 a , ICML Workshop: Sampling and Optimization in Discrete Space

    Adachi, M., Hayakawa, S., Hamid, S., et al. 2023 a , ICML Workshop: Sampling and Optimization in Discrete Space

    2023

  2. [2]

    Adachi, M., Hayakawa, S., J rgensen, M., Oberhauser, H., & Osborne, M. A. 2022, in Advances in Neural Information Processing Systems, ed. S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, & A. Oh, Vol. 35 (Curran Associates, Inc.), 16533--16547

    2022

  3. [3]

    2024 a , https://doi.org/10.48550/arXiv.2404.12219

    Adachi, M., Hayakawa, S., J rgensen, M., et al. 2024 a , https://doi.org/10.48550/arXiv.2404.12219

    arXiv doi:10.48550/arxiv.2404.12219 2024

  4. [4]

    2024 b , in International Conference on Artificial Intelligence and Statistics, Vol

    Adachi, M., Hayakawa, S., J rgensen, M., et al. 2024 b , in International Conference on Artificial Intelligence and Statistics, Vol. 238, PMLR, 496--504

    2024

  5. [5]

    2023 b , IFAC-PapersOnLine, 56, 10521

    Adachi, M., Kuhn, Y., Horstmann, B., et al. 2023 b , IFAC-PapersOnLine, 56, 10521

    2023

  6. [6]

    2024 c , in International Conference on Artificial Intelligence and Statistics, PMLR, 505--513

    Adachi, M., Planden, B., Howey, D., et al. 2024 c , in International Conference on Artificial Intelligence and Statistics, PMLR, 505--513

    2024

  7. [7]

    2024, , 687, A311

    Alaguero , A., Cuello , N., M \'e nard , F., et al. 2024, , 687, A311

    2024

  8. [8]

    T., M \"u ller , H

    Belloche , A., Garrod , R. T., M \"u ller , H. S. P., & Menten , K. M. 2014, Science, 345, 1584

    2014

Show all 98 references
  1. [9]

    Belloche , A., M \"u ller , H. S. P., Garrod , R. T., & Menten , K. M. 2016, , 587, A91

    2016

  2. [10]

    Belloche , A., M \"u ller , H. S. P., Menten , K. M., Schilke , P., & Comito , C. 2013, , 559, A47

    2013

  3. [11]

    2019, , 483, 1850

    Bianchi , E., Codella , C., Ceccarelli , C., et al. 2019, , 483, 1850

    2019

  4. [12]

    2023, Faraday Discussions, 245, 164

    Bianchi , E., L \'o pez-Sepulcre , A., Ceccarelli , C., et al. 2023, Faraday Discussions, 245, 164

    2023

  5. [13]

    P., Jankowiak, M., et al

    Bingham, E., Chen, J. P., Jankowiak, M., et al. 2019, J. Mach. Learn. Res., 20, 28:1

    2019

  6. [14]

    M., Kucukelbir , A., & McAuliffe , J

    Blei , D. M., Kucukelbir , A., & McAuliffe , J. D. 2016, arXiv e-prints, arXiv:1601.00670

    2016 arXiv

  7. [15]

    S., Walsh , C., Terwisscha van Scheltinga , J., et al

    Booth , A. S., Walsh , C., Terwisscha van Scheltinga , J., et al. 2021, Nature Astronomy, 5, 684

    2021

  8. [16]

    A., Belloche , A., Garrod , R

    Busch , L. A., Belloche , A., Garrod , R. T., M \"u ller , H. S. P., & Menten , K. M. 2024, , 681, A104

    2024

  9. [17]

    & Ceccarelli , C

    Caselli , P. & Ceccarelli , C. 2012, , 20, 56

    2012

  10. [18]

    M., Zucconi , A., et al

    Caselli , P., Walmsley , C. M., Zucconi , A., et al. 2002, , 565, 344

    2002

  11. [19]

    2014, in Protostars and Planets VI, ed

    Ceccarelli , C., Caselli , P., Bockel \'e e-Morvan , D., et al. 2014, in Protostars and Planets VI, ed. H. Beuther , R. S. Klessen , C. P. Dullemond , & T. Henning , 859--882

    2014

  12. [20]

    Ceccarelli , C., Caselli , P., Herbst , E., Tielens , A. G. G. M., & Caux , E. 2007, in Protostars and Planets V, ed. B. Reipurth , D. Jewitt , & K. Keil , 47

    2007

  13. [21]

    2023, in Astronomical Society of the Pacific Conference Series, Vol

    Ceccarelli , C., Codella , C., Balucani , N., et al. 2023, in Astronomical Society of the Pacific Conference Series, Vol. 534, Protostars and Planets VII, ed. S. Inutsuka , Y. Aikawa , T. Muto , K. Tomida , & M. Tamura , 379

    2023

  14. [22]

    2022, , 657, A78

    Chahine , L., L \'o pez-Sepulcre , A., Neri , R., et al. 2022, , 657, A78

    2022

  15. [23]

    2022, , 515, 543

    Codella , C., L \'o pez-Sepulcre , A., Ohashi , S., et al. 2022, , 515, 543

    2022

  16. [24]

    C., Boulanger , A., et al

    Coutens , A., Loison , J. C., Boulanger , A., et al. 2022, , 660, L6

    2022

  17. [25]

    V., J rgensen , J

    Coutens , A., Persson , M. V., J rgensen , J. K., Wampfler , S. F., & Lykke , J. M. 2015, , 576, A5

    2015

  18. [26]

    M., Mandel , K

    Czekala , I., Andrews , S. M., Mandel , K. S., Hogg , D. W., & Green , G. M. 2015, , 812, 128

    2015

  19. [27]

    2017, arXiv e-prints, arXiv:1708.00955

    Dang , K.-D., Quiroz , M., Kohn , R., Tran , M.-N., & Villani , M. 2017, arXiv e-prints, arXiv:1708.00955

    2017 arXiv

  20. [28]

    2014, arXiv e-prints, arXiv:1410.8516

    Dinh , L., Krueger , D., & Bengio , Y. 2014, arXiv e-prints, arXiv:1410.8516

    2014 arXiv

  21. [29]

    N., Dishoeck, E

    Drozdovskaya, M. N., Dishoeck, E. F. v., Rubin, M., Jørgensen, J. K., & Altwegg, K. 2019, Monthly Notices of the Royal Astronomical Society, 490, 50

    2019

  22. [30]

    2014, PhD thesis, University of Cambridge

    Duvenaud, D. 2014, PhD thesis, University of Cambridge

    2014

  23. [31]

    P., Schlemmer , S., Schilke , P., Stutzki , J., & M \"u ller , H

    Endres , C. P., Schlemmer , S., Schilke , P., Stutzki , J., & M \"u ller , H. S. P. 2016, Journal of Molecular Spectroscopy, 327, 95

    2016

  24. [32]

    Frank , P., Leike , R., & En lin , T. A. 2021, Entropy, 23, 853

    2021

  25. [33]

    J., & Leary, C

    Frostig, R., Johnson, M. J., & Leary, C. 2018, Systems for Machine Learning, 4

    2018

  26. [34]

    Q., Bindel, D., & Wilson, A

    Gardner, J., Pleiss, G., Weinberger, K. Q., Bindel, D., & Wilson, A. G. 2018, Advances in neural information processing systems, 31

    2018

  27. [35]

    T., Widicus Weaver , S

    Garrod , R. T., Widicus Weaver , S. L., & Herbst , E. 2008, , 682, 283

    2008

  28. [36]

    & Domke, J

    Geffner, T. & Domke, J. 2021, in Advances in Neural Information Processing Systems, ed. M. Ranzato, A. Beygelzimer, Y. Dauphin, P. Liang, & J. W. Vaughan, Vol. 34 (Curran Associates, Inc.), 639--651

    2021

  29. [37]

    2022, , 163, 291

    Ginsburg , A., Sokolov , V., de Val-Borro , M., et al. 2022, , 163, 291

    2022

  30. [38]

    & Neuhoff, D

    Gray, R. & Neuhoff, D. 1998, IEEE Transactions on Information Theory, 44, 2325

    1998

  31. [39]

    & Barber, D

    Habib, R. & Barber, D. 2019, in International Conference on Learning Representations

    2019

  32. [40]

    R., & Wardle, M

    Hafner, A., Dawson, J. R., & Wardle, M. 2021, The Astrophysical Journal, 923, 261

    2021

  33. [41]

    2022, Advances in Neural Information Processing Systems, 35, 6886

    Hayakawa, S., Oberhauser, H., & Lyons, T. 2022, Advances in Neural Information Processing Systems, 35, 6886

    2022

  34. [42]

    A., & Kersting, H

    Hennig, P., Osborne, M. A., & Kersting, H. P. 2022, Probabilistic Numerics: Computation as Machine Learning (Cambridge University Press)

    2022

  35. [43]

    D., Longmore , S

    Henshaw , J. D., Longmore , S. N., Kruijssen , J. M. D., et al. 2016, , 457, 2675

    2016

  36. [44]

    & van Dishoeck , E

    Herbst , E. & van Dishoeck , E. F. 2009, , 47, 427

    2009

  37. [45]

    M., Wang , C., & Paisley , J

    Hoffman , M., Blei , D. M., Wang , C., & Paisley , J. 2012, arXiv e-prints, arXiv:1206.7051

    2012 arXiv

  38. [46]

    & Ma, Y.-A

    Hoffman, M. & Ma, Y.-A. 2020, in Proceedings of the 37th International Conference on Machine Learning, ICML'20 (JMLR.org)

    2020

  39. [47]

    Hoffman , M. D. & Gelman , A. 2011, arXiv e-prints, arXiv:1111.4246

    2011 arXiv

  40. [49]

    & Mohamed , S

    Jimenez Rezende , D. & Mohamed , S. 2015 b , arXiv e-prints, arXiv:1505.05770

    2015 arXiv

  41. [50]

    M., et al

    Jim \'e nez-Serra , I., Mart \' n-Pintado , J., Rivilla , V. M., et al. 2020, Astrobiology, 20, 1048

    2020

  42. [51]

    K., Belloche , A., & Garrod , R

    J rgensen , J. K., Belloche , A., & Garrod , R. T. 2020, , 58, 727

    2020

  43. [52]

    K., van der Wiel , M

    J rgensen , J. K., van der Wiel , M. H. D., Coutens , A., et al. 2016, , 595, A117

    2016

  44. [53]

    2024, A&A, 685, A164

    Juvela & Tharakkal . 2024, A&A, 685, A164

    2024

  45. [54]

    2018, in Proceedings of Machine Learning Research, Vol

    Kandasamy, K., Krishnamurthy, A., Schneider, J., & Poczos, B. 2018, in Proceedings of Machine Learning Research, Vol. 84, Proceedings of the Twenty-First International Conference on Artificial Intelligence and Statistics, ed. A. Storkey & F. Perez-Cruz (PMLR), 133--142

    2018

  46. [55]

    2017, , 469, L73

    Lefloch , B., Ceccarelli , C., Codella , C., et al. 2017, , 469, L73

    2017

  47. [56]

    R., Vera-Ciro , C., Murray , C

    Lindner , R. R., Vera-Ciro , C., Murray , C. E., et al. 2015, , 149, 138

    2015

  48. [57]

    A., \"O berg , K

    Loomis , R. A., \"O berg , K. I., Andrews , S. M., et al. 2018, , 155, 182

    2018

  49. [58]

    & Hutter , F

    Loshchilov , I. & Hutter , F. 2016, arXiv e-prints, arXiv:1608.03983

    2016 arXiv

  50. [59]

    J., Balandat, M., Wilson, A

    Maddox, W. J., Balandat, M., Wilson, A. G., & Bakshy, E. 2021, Advances in neural information processing systems, 34, 19274

    2021

  51. [60]

    K., Calcutt , H., et al

    Manigand , S., J rgensen , J. K., Calcutt , H., et al. 2020, , 635, A48

    2020

  52. [61]

    2019, A&A, 626, A101

    Marchal , Miville-Deschênes, Marc-Antoine , Orieux, François , et al. 2019, A&A, 626, A101

    2019

  53. [62]

    , Blanco-Sánchez, C

    Martín , Martín-Pintado, J. , Blanco-Sánchez, C. , et al. 2019, A&A, 631, A159

    2019

  54. [63]

    A., Burkhardt , A

    McGuire , B. A., Burkhardt , A. M., Kalenskii , S., et al. 2018, Science, 359, 202

    2018

  55. [64]

    Meech , A., Aigrain , S., Brogi , M., & Birkby , J. L. 2022, , 512, 2604

    2022

  56. [65]

    N., Savage , C., Brewster , M

    Milam , S. N., Savage , C., Brewster , M. A., Ziurys , L. M., & Wyckoff , S. 2005, , 634, 1126

    2005

  57. [66]

    , & Schilke, P

    Möller , Endres, C. , & Schilke, P. 2017, A&A, 598, A7

    2017

  58. [67]

    1991, Journal of Statistical Planning and Inference, 29, 245

    O'Hagan, A. 1991, Journal of Statistical Planning and Inference, 29, 245

    1991

  59. [68]

    2012, in Advances in Neural Information Processing Systems (NeurIPS), Vol

    Osborne, M., Garnett, R., Ghahramani, Z., et al. 2012, in Advances in Neural Information Processing Systems (NeurIPS), Vol. 25

    2012

  60. [69]

    & Sandberg, I

    Park, J. & Sandberg, I. W. 1991, Neural computation, 3, 246

    1991

  61. [70]

    2019, arXiv preprint arXiv:1912.11554

    Phan, D., Pradhan, N., & Jankowiak, M. 2019, arXiv preprint arXiv:1912.11554

    2019 arXiv

  62. [71]

    E., Arzoumanian , D., Andre , P., et al

    Pineda , J. E., Arzoumanian , D., Andre , P., et al. 2023, in Astronomical Society of the Pacific Conference Series, Vol. 534, Protostars and Planets VII, ed. S. Inutsuka , Y. Aikawa , T. Muto , K. Tomida , & M. Tamura , 233

    2023

  63. [72]

    E., Maury , A

    Pineda , J. E., Maury , A. J., Fuller , G. A., et al. 2012, , 544, L7

    2012

  64. [73]

    Ranganath , R., Gerrish , S., & Blei , D. M. 2013, arXiv e-prints, arXiv:1401.0118

    2013 arXiv

  65. [74]

    Rasmussen, C. E. & Ghahramani, Z. 2002, in Advances in Neural Information Processing Systems (NeurIPS), Vol. 15, 505--512

    2002

  66. [75]

    Rasmussen, C. E. & Williams, C. K. I. 2005, Gaussian Processes for Machine Learning (The MIT Press)

    2005

  67. [76]

    , Henshaw, J

    Riener , Kainulainen, J. , Henshaw, J. D. , et al. 2019, A&A, 628, A78

    2019

  68. [77]

    M., Colzi , L., Jim \'e nez-Serra , I., et al

    Rivilla , V. M., Colzi , L., Jim \'e nez-Serra , I., et al. 2022, , 929, L11

    2022

  69. [78]

    Rodgers , S. D. & Millar , T. J. 1996, , 280, 1046

    1996

  70. [79]

    P., & Welling , M

    Salimans , T., Kingma , D. P., & Welling , M. 2014, arXiv e-prints, arXiv:1410.6460

    2014 arXiv

  71. [80]

    E., Buchner , J., & Caselli , P

    Sokolov , V., Pineda , J. E., Buchner , J., & Caselli , P. 2020, , 892, L32

    2020

  72. [81]

    M., & Seeger, M

    Srinivas, N., Krause, A., Kakade, S. M., & Seeger, M. 2009, International Conference on Machine Learning

    2009

  73. [82]

    Stein, M. L. 1999, Interpolation of spatial data (Springer Science & Business Media)

    1999

  74. [83]

    van Dishoeck , E. F. & Blake , G. A. 1998, , 36, 317

    1998

  75. [84]

    Vasyunin , A. I. & Herbst , E. 2013, , 769, 34

    2013

  76. [85]

    & Weber , T

    Wingate , D. & Weber , T. 2013, arXiv e-prints, arXiv:1301.1299

    2013 arXiv

  77. [86]

    N., & Osborne, M

    Xu, W., Adachi, M., Jones, C. N., & Osborne, M. A. 2024, arXiv preprint arXiv:2410.10452

    2024 arXiv

  78. [87]

    2017, arXiv e-prints, arXiv:1711.05597

    Zhang , C., Butepage , J., Kjellstrom , H., & Mandt , S. 2017, arXiv e-prints, arXiv:1711.05597

    2017 arXiv

  79. [88]

    1997, ACM Transactions on Mathematical Software, 23, 550

    Zhu, C., Byrd, R., Lu, P., & Nocedal, J. 1997, ACM Transactions on Mathematical Software, 23, 550

    1997

  80. [89]

    1966, in Stellar Evolution, ed.\ R

    Baker, N. 1966, in Stellar Evolution, ed.\ R. F. Stein,& A. G. W. Cameron (Plenum, New York) 333

    1966

  81. [90]

    1988, A&A, 200, 58

    Balluch, M. 1988, A&A, 200, 58

    1988

  82. [91]

    Cox, J. P. 1980, Theory of Stellar Pulsation (Princeton University Press, Princeton) 165

    1980

  83. [92]

    N.,& Stewart, J

    Cox, A. N.,& Stewart, J. N. 1969, Academia Nauk, Scientific Information 15, 1

    1969

  84. [93]

    1980, Prog

    Mizuno H. 1980, Prog. Theor. Phys., 64, 544

    1980

  85. [94]

    Tscharnuter W. M. 1987, A&A, 188, 55

    1987

  86. [95]

    1992, in ASP Conf

    Terlevich, R. 1992, in ASP Conf. Ser. 31, Relationships between Active Galactic Nuclei and Starburst Galaxies, ed. A. V. Filippenko, 13

    1992

  87. [96]

    Yorke, H. W. 1980a, A&A, 86, 286

  88. [97]

    F., Tytler, D

    Zheng, W., Davidsen, A. F., Tytler, D. & Kriss, G. A. 1997, preprint

    1997

  89. [98]

    , " * write output.state after.block = add.period write newline

    ENTRY address archiveprefix author booktitle chapter edition editor howpublished institution eprint journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sent...

  90. [99]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.