betaFeedback

WorksPhysics and Astronomy

On the reduction of negative-weight events in MC@NLO with Machine Learning techniques

Abstract

We demonstrate that Machine Learning techniques can be successfully employed to re-define S{\mathbb S}-event MC@NLO short-distance cross sections in a manner which in most cases leads to a reduction of their fractions of negative weights, with a modest increase of computational costs. We work with both Decision Trees and Normalizing Flows, and consider several processes, in both e+e−e^+e^- and pppp collisions, that vastly differ in their kinematical and dynamical complexity. In those situations where the ML-induced reduction of negative weights is not significant we are able to identify the primary reason(s) for this behaviour, thus paving the way for it to be addressed in future studies. This work is still to be considered as a proof of concept, given that the proposed methodology is not implemented in a public code; yet, it is based upon a fully realistic simulation environment, in which no simplifying assumptions are made.

Introduction

The spectacular performances of the LHC and of the experiments which observe the results of the collisions that happen there have turned high-energy hadronic physics into a field of high-precision measurements, to an extent that was difficult to anticipate at the beginning of its operations. Most of the dominant uncertainties affecting data are of theoretical origin, and thus the necessity of increasing the accuracy of predictions is accepted as a matter of fact by the theory community. Part of that effort concerns tools which are employed by the experiments themselves, namely the Monte Carlo (MC) programs that can simulate hard collisions and their accompanying phenomena in minute details.

While the majority of the number of events simulated by the LHC collaborations are still based on an underlying Leading Order (LO) description, such a dominance is not overwhelming as it used to be, since Next-to-Leading Order (NLO) accurate simulations are nowadays considered as standard, and occupy a significant part of the computing resources allocated by experiments, in keeping with the sought increase of precision of the predictions.

Unfortunately, NLO simulations have a number of issues which do not affect their LO counterparts. Namely, they are slower and more complicated to run, they can reach smaller hard-event-level final-state multiplicities, and feature events with negative weights, which effectively reduce the statistical power of their predictions.

This is the reason why the reduction of negative weights in NLO simulations has been a constant topic of discussion for the past twenty-five years, after the invention of MC@NLO [1]; it is both a conceptually interesting problem, and a very practical issue as well, since such weights cost real money. At the level of formalisms, the most developed (in terms of both corollary theoretical ideas and applications) is Powheg [2, 3, 4]; other proposals [5, 6, 7, 8, 9, 10] are more recent and studied in a much less systematic manner. A different viewpoint is that which informs the works aimed at solving the problem starting from an existing NLO-accurate method (typically, MC@NLO), as is done e.g. in refs. [11, 12, 13, 14, 15, 16, 17, 18], or by means of a-posteriori manipulations of both-sign event files [19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30].

The present work belongs to the former category of the two we have just mentioned; more specifically, the idea is that of modifying the MC@NLO short-distance S-event cross section in a local way, so that the number of negative weights is reduced without affecting physical predictions. While our work can be seen as a generalization of the Born-spreading technique of ref. [13], its defining characteristics lies in the fact that the modifications mentioned before are entirely determined by means of Machine Learning (ML) techniques; in particular, we employ both Decision Trees (DT) and Normalizing Flows (NF). The main reason for selecting such approaches lies in their suitability for efficiently evaluating integrals of the learned function over the subspace of radiation variables, as is required by the procedure we are going to define. We stress again that the method deals only with SS events, and that HH events are not relevant in this context. Having said that, it is interesting to observe that several of the papers cited above focus on HH events instead. We shall in fact explicitly consider some MC@NLO-Δ\Delta [11] simulations, which by construction reduce the number of negative HH events, “moving” some of them into the SS-event set, where they can be handled with the procedures we present in this work, thereby showing a synergy between the two approaches.

This paper is organised as follows: we start in sect. 2 with an elementary introduction to the basic ML concepts we shall use in this paper. In sect. 3 we set up the notation, and write the MC@NLO and MC@NLO-Δ\Delta cross sections accordingly. We introduce the ML-driven method we propose in this work in sect. 3.4, with DT- and NF-specific strategies discussed in sects. 3.4.1 and 3.4.2, respectively. In sect. 4 we define a metric which we use to assess whether our ML-based procedure is viable, in that it provides one with benefits superior to its overall cost. We then document in sect. 5 how the procedure performs in the context of actual e+e−e^{+}e^{-}- and pppp-collisions processes. The DT- and NF-induced results are discussed separately, in sects. 5.1 and 5.2 for the former, and in sects. 5.3 and 5.4 for the latter. We point out that we identify general issues outside of scope of this work, that are independent of which ML approach one adopts, and which we discuss within the DT sections. This by no means implies that we consider DTs superior to NFs for the problem at hand. Our overall conclusions and outlook are presented in sect. 6; these are mostly non-technical, since DT- and NF-specific summaries are given in sect. 5.2.11 and 5.4.3, respectively. Additional material on DTs and NFs can be found in the appendices.

Introduction to machine learning

Machine Learning (ML) is a collection of computational methods that construct models from data, rather than relying exclusively on explicitly programmed rules. The central idea is to approximate an unknown function or probability distribution using a parametrized model whose parameters (called trainable) are determined through an optimization procedure.

Suppose we have a dataset

D={(xi,yi)}i=1N,(1)\mathcal{D} = \left\{(x_i,y_i)\right\}_{i=1}^{N}, \tag*{(1)}

where xix_i denotes a set of input variables and yiy_i a corresponding target. We introduce a parametrized function fθ(x)f_{\theta}(x), where θ\theta represents the model trainable parameters. The objective is to determine θ\theta so that fθ(xi)f_{\theta}(x_i) approximates yiy_i. This is typically achieved by minimizing a loss function:

θ∗=argmin⁡θ L(θ).(2)\theta^{*} = \underset{\theta}{\operatorname{argmin}}\,\mathcal{L}(\theta). \tag*{(2)}

For example, in regression1 one may choose the mean squared error,

L(θ)=1N∑i=1N[fθ(xi)−yi]2.(3)\mathcal{L}(\theta) = \frac{1}{N}\sum_{i=1}^{N} [f_\theta(x_i) - y_i]^2 . \tag*{(3)}

A crucial requirement is that the resulting model perform well on previously unseen data, rather than merely reproducing its training sample. This property is called generalization.

In addition to the trainable parameters, a model is characterized by the so-called hyperparameters. These specify the model structure, and/or how training is performed.

There are several broad classes of ML methods. In supervised learning, the training data contain known targets; typical tasks are regression and classification. In unsupervised learning, the algorithm identifies patterns or structures in data without prescribed targets. Generative modelling aims to learn a probability distribution and generate new samples consistent with it. These categories overlap: in particular, generative models can be trained using supervised or unsupervised objectives.

Common ML algorithms include decision trees, which partition the input space into regions; Boosted Decision Trees, which train successive trees adaptively to improve the combined prediction; neural networks, which construct flexible functions through compositions of parametrized transformations; and Normalizing Flows, which learn probability densities through invertible transformations.

For the negative-weight reduction problem in MC@NLO simulations, ML will be used to construct a correction to an event-weight function. Schematically, denoting by w(ρ,x)w(\rho, x) that function, with ρ\rho and xx two independent (sets of) variables, the correction will amount to the replacement:

w(ρ,x)⟶w′(ρ,x)=w(ρ,x)+fθ(ρ,x),(4)w(\rho, x) \longrightarrow w'(\rho, x) = w(\rho, x) + f_\theta(\rho, x), \tag*{(4)}

where the learned function fθ(ρ,x)f_\theta(\rho, x) is constrained so that

∫dx fθ(ρ,x)=0for every fixed ρ.(5)\int dx\, f_\theta(\rho, x) = 0 \qquad\text{for every fixed }\rho. \tag*{(5)}

The objective is to reduce the fraction of negative-weight events without modifying the physical cross section and relevant differential distributions. Integrals analogous to that in eq. (5) need to be computed on the fly (i.e. very efficiently) in our procedure, which is what has been alluded to in the introduction when discussing the motivations beyond our choices of DT and NF as ML tools. For example, while the present regression task could be also handled by unconstrained Neural Networks, the computation of an integral with high efficiency would not.

Decision Trees

Decision Trees [31, 32, 33] approximate a function by recursively partitioning the space of its arguments. We use trees for regression: the prediction is a real-valued approximation to the chosen target, and not a classification of events according to the sign of their weights.

A decision tree divides the space of input variables into progressively smaller regions by asking a sequence of simple questions. For example, if the input variables describe the kinematics of a particle collision, a tree might first separate events according to whether their energy exceeds a given threshold, and then further subdivide them according to their angles or transverse momenta. Each final region in the input-variable space, called a leaf, is assigned a numerical prediction. In this way, a decision tree approximates a potentially complicated function by a collection of simpler, locally constant functions.

For learning inputs zz, a constant-leaf regression tree has the representation [34]

T(z)=∑ℓ=1Laℓ1Rℓ(z),(6)T(z) = \sum_{\ell=1}^{L} a_{\ell}\mathbf{1}_{R_{\ell}}(z), \tag*{(6)}

where RℓR_{\ell} is the region associated with leaf ℓ\ell, aℓa_{\ell} is its prediction, and LL is the number of leaves; the quantity 1Rℓ(z)\mathbf{1}_{R_{\ell}}(z) is equal to one if z∈Rℓz \in R_{\ell}, and zero otherwise. With axis-aligned splits, the regions RℓR_{\ell} are hyperrectangles in the learning variables; their geometry is discussed in appendix A.§ 2.

In the general ML notation, a tree defines a parametrized function fθ(z)f_{\theta}(z), where θ\theta specifies its split variables, thresholds and leaf predictions. The tree is constructed by a sequence of greedy split optimizations2, not by a gradient-descent training loop. For squared-error regression, such as the one predominantly used in the DT applications in this work, the prediction in each leaf is the mean of the training targets reaching that leaf, using the supplied sample weights when present [34]. After training, the tree is held fixed during integration and event generation.

An important consideration is the balance between accuracy and complexity. Trees with very fine subdivisions can describe the training data extremely accurately, but may also reproduce statistical fluctuations that are not representative of the underlying distribution. This phenomenon, known as overfitting, can be controlled by limiting tree complexity, adjusting the minimum number of events assigned to each leaf, and evaluating the model on independent data.

In the context of complex Monte Carlo simulations, DTs constitute an attractive choice owing to their ability to learn intricate dependences on several kinematic variables, without requiring an explicit analytical model of the corrections. Their effectiveness ultimately depends on whether the information contained in the training variables is sufficient to identify the structures which one seeks to reproduce.

Training determines the split variables, their thresholds, and the predictions assigned to leaves. The principal settings chosen in advance (i.e. the model hyperparameters) include [35].3

  • Minimum samples per leaf (nmin⁡n_{\min}, min_samples_leaf): the smallest number of training events allowed in a final region; this is the principal granularity control varied in our DT study.

  • Maximum tree depth (dmax⁡d_{\max}, max_depth): the maximum number of successive splits along a root-to-leaf path.

  • Maximum number of leaves (max_leaf_nodes): a bound on the total number of terminal regions.

  • Minimum samples for a split (min_samples_split): the minimum node occupancy required before a split is considered.

  • Split criterion and pruning controls: the criterion used to select subdivisions and any controls that remove unnecessary branches.

The numerical settings we have used are given in section 5.2.

Extensions to tree ensembles. Decision trees can also be combined into ensembles. In Boosted Decision Trees (BDTs), successive trees are trained using information from the preceding predictions. For gradient boosting with squared-error regression, each additional tree is fitted to the residuals of the current ensemble [36]:

Fm(z)=Fm−1(z)+ηBDTTm(z),ri(m)=yi−Fm−1(zi).(7)\begin{aligned} F_m(z) &= F_{m-1}(z) + \eta_{\mathrm{BDT}}T_m(z),\\ r_i^{(m)} &= y_i - F_{m-1}(z_i). \tag*{(7)} \end{aligned}

Here F0F_0 is an initial prediction, TmT_m is the tree fitted at stage mm, and ηBDT\eta_{\mathrm{BDT}} is the learning rate. This is a specific gradient-boosting example, rather than a definition of every boosting algorithm. The sequential dependence concerns training: the final prediction is an additive combination of trees evaluated on the same input, not a composition of their outputs.

The structural hyperparameters retain their meaning for each constituent tree, while the number of boosting stages, learning rate and possible subsampling introduce additional controls. Random forests instead average predictions from randomized trees [36]; these ensembles are possible extensions of the present DT implementation. Their further use should be assessed together with their training, evaluation and integration costs through the figure of merit of section 4. The radiation integral of an additive ensemble is discussed in appendix A.3.

Normalizing Flows

Normalizing Flows [37, 38, 39, 40, 41] are a class of ML models designed to describe probability distributions. While many conventional ML algorithms aim to predict a numerical quantity from a set of input variables, Normalizing Flows focus on learning how probability is distributed over a potentially complicated multidimensional space. This makes them particularly useful in problems involving event generation, statistical sampling, and Monte Carlo integration.

The central idea is to start from a simple, known probability distribution, and transform it into a more complicated one through a sequence of reversible mathematical operations. For example, one may start from points uniformly distributed inside a multidimensional unit cube. A Normalizing Flow progressively transforms these points so that they become more concentrated in some regions and less concentrated in others, according to a probability distribution learned from the data. In general, Normalizing Flows are characterized by their flexibility in describing complicated distributions, their ability to generate new samples efficiently, and the normalization properties of their learned densities.

Each transformation is constructed so that it can be inverted and its effect on probability densities can be computed. This is an essential feature: it allows the model not only to generate points according to the learned distribution, but also to evaluate the corresponding probability density. Furthermore, since the starting distribution is normalized and the transformations correctly account for changes in volume, the resulting probability distribution is automatically normalized.

In keeping with the general ML framework, the transformations depend on a set of parameters θ\theta, whose values are determined through training. The resulting model is a parametrized probability density, which we may denote by fθ(z)f_\theta(z). A loss function L(θ)\mathcal{L}(\theta) specifies the desired properties of this density, and training determines the parameters that optimize it. Depending on the application, the objective may be to reproduce a given distribution or to construct one that improves the efficiency of a numerical procedure.

A particularly useful extension is the conditional Normalizing Flow, in which the learned probability distribution depends on additional input variables. In our case, for example, we shall construct a distribution of radiation variables whose shape depends on the underlying Born kinematics. Thus, the same model can describe different radiation patterns for different Born configurations.

Training determines the weights within the internal neural networks. Typical hyperparameter settings include:

  • Number and type of transformations (NtransN_{\mathrm{trans}}, T\mathcal{T}): how many invertible steps are used (e.g. affine or spline-based).

  • Spline bins (for spline-based flows): how flexibly each step can adapt its shape.

  • Network size (LL, nunitsn_{\mathrm{units}}): the number of internal layers and units per layer.

  • Training settings: learning rate (ηNF\eta_{\mathrm{NF}}), batch size (nbatchn_{\mathrm{batch}}), and number of epochs (NepochN_{\mathrm{epoch}}).

The parameters which appear in the last item are those that are easiest to handle on an individual-run basis, and therefore we elaborate further on them.

The learning rate ηNF\eta_{\mathrm{NF}} controls the scale of each update of the trainable parameters of the flows and their auxiliary networks. For ordinary gradient descent, for example, the update4 is θ→θ−ηNF∇θL(θ)\theta\to\theta-\eta_{\mathrm{NF}}\nabla_\theta\mathcal{L}(\theta); other optimizers modify the update direction while retaining a learning-rate scale. Too large a learning rate can destabilise training, whereas too small a value can make progress unnecessarily slow.

The batch size nbatchn_{\mathrm{batch}} specifies how many configurations are processed together to evaluate the loss and its gradients for one parameter update. When this number is smaller than the training sample, these groups are called mini-batches. The parameters are updated after each batch, rather than only after the complete sample has been processed. The actual size of the last batch may be smaller than nbatchn_{\mathrm{batch}}.

For a fixed training sample, an epoch is one complete pass through that sample. The number of epochs NepochN_{\mathrm{epoch}} therefore counts repeated passes over the same configurations, typically with their order reshuffled between passes; it does not count independent new samples. With one update per batch and no discarded configurations, the number of updates is

Nupdate=Nepoch⌊Ntrainnbatch⌋.(8)N_{\mathrm{update}} = N_{\mathrm{epoch}} \left\lfloor\frac{N_{\mathrm{train}}}{n_{\mathrm{batch}}} \right\rfloor. \tag*{(8)}

Thus changing the batch size also changes the number of updates at a fixed number of epochs.

MC@NLO definitions and learning functions

In this section, we first collect the basic MC@NLO and MC@NLO-Δ\Delta formulae, and then discuss how these can be employed in the context of an ML-driven negative-weight reduction procedure. These formulae are understood for a fixed real-emission subprocess and FKS pair, with the physical sums over these labels implicit. Also, it will soon become apparent that the applied ML methodology is identical for MC@NLO and MC@NLO-Δ\Delta; thus, the latter will be generally understood.

Phase-space variables

Henceforth, we consider a process which, at the Born level, features nn final-state particles. By

χn,χr,(9)\chi_n,\qquad\chi_r, \tag*{(9)}

we collectively denote, respectively, the 3n−43n-4 phase-space variables at the Born level, and the three variables relevant to the radiation of an extra parton at the real-emission level - the latter may be equivalently called radiation or FKS variables. The (n+1)(n+1)-body phase space is written as follows:

dϕn+1=Φ(n+1)(χn+1) dχn+1≡Φ(n+1)(χn,χr) dχndχr,(10)d\phi_{n+1} = \Phi^{(n+1)}(\chi_{n+1})\,d\chi_{n+1} \equiv\Phi^{(n+1)}(\chi_n,\chi_r)\,d\chi_n d\chi_r, \tag*{(10)}
χn+1=χn∪χr,(11)\chi_{n+1} = \chi_n \cup\chi_r, \tag*{(11)}
χr={ξ,y,φ}.(12)\chi_r = \{\xi,y,\varphi\}. \tag*{(12)}

In the FKS parametrisation (which uses the incoming-parton c.m. frame as a reference), ξ\xi is the rescaled energy of the FKS parton, yy the cosine of the angle between that parton and its sister, and φ\varphi an azimuthal angle. The soft and collinear limits are ξ→0\xi\to0 and y→1y \to1, respectively. Furthermore

0≤ξ≤ξmax⁡(χn),−1≤y≤1,0≤φ≤2π.(13)0 \le\xi\le\xi_{\max}(\chi_n), \qquad-1 \le y \le1, \qquad0 \le\varphi\le2\pi. \tag*{(13)}

It is then convenient to introduce the rescaled variables

ξ^=ξξmax⁡(χn),y^=12(1−y),φ^=φ2π,(14)\hat{\xi} = \frac{\xi}{\xi_{\max}(\chi_n)}, \qquad\hat{y} = \frac{1}{2}(1-y), \qquad\hat{\varphi} = \frac{\varphi}{2\pi}, \tag*{(14)}

so that the soft and collinear limits are ξ^→0\hat{\xi} \to0 and y^→0\hat{y} \to0, respectively, and importance sampling towards those limits can be achieved by means of a power law ξ^k\hat{\xi}^{k}, y^k\hat{y}^{k}, with k>1k > 1.

This parametrisation is valid in each FKS sector; we remind the reader that FKS sectors (and thereby the FKS variables relevant to each of them) are identified by a pair of integer indices (see footnote 16 for their physical interpretation); for the sake of a simpler notation, these will be systematically omitted in what follows. The two additional Bjorken-xx integration variables will also be understood.

Event configurations with real-emission (H events) and Born kinematics (S events) will be denoted by K(H)\mathcal{K}^{(\mathrm{H})} and K(S)\mathcal{K}^{(\mathrm{S})}, respectively. They are such that:

K(S)=K(S)(χn),K(H)=K(H)(χn,χr).(15)\mathcal{K}^{(\mathrm{S})} = \mathcal{K}^{(\mathrm{S})}(\chi_n), \qquad\mathcal{K}^{(\mathrm{H})} = \mathcal{K}^{(\mathrm{H})}(\chi_n,\chi_r). \tag*{(15)}

We shall often denote a fully differential cross section times its relevant infinitesimal measure by a shorthand notation such as dσd\sigma. In a more explicit manner, this implies that a function ς\varsigma exists, so that

dσ=ς(χn,χr)dχndχr=ς(χn,χr)Φ(n+1)(χn,χr)dϕn+1.(16)d\sigma= \varsigma(\chi_n,\chi_r)d\chi_n d\chi_r = \frac{\varsigma(\chi_n,\chi_r)}{\Phi^{(n+1)}(\chi_n,\chi_r)}d\phi_{n+1}. \tag*{(16)}

The function ς\varsigma is then the cross section fully differential in (χn,χr)(\chi_n,\chi_r), or can be used to construct a Born-level differential quantity:

References

  1. [1]S. Frixione and B. R. Webber, Matching NLO QCD computations and parton shower simulations, JHEP 06 (2002) 029, [hep-ph/0204244].DOI
  2. [2]P. Nason, A New Method for Combining NLO QCD with Shower Monte Carlo Algorithms, JHEP 11 (2004) 040, [hep-ph/0409146].DOI
  3. [3]S. Frixione, P. Nason and C. Oleari, Matching NLO QCD computations with Parton Shower simulations: the POWHEG method, JHEP 11 (2007) 070, [0709.2092].DOI
  4. [4]S. Alioli, P. Nason, C. Oleari and E. Re, A general framework for implementing NLO calculations in shower Monte Carlo programs: the POWHEG BOX, JHEP 06 (2010) 043, [1002.2581].DOI
  5. [5]S. Jadach, W. Placzek, S. Sapeta, A. Siód­mok and M. Skrzypek, Matching NLO QCD with parton shower in Monte Carlo scheme—the KrkNLO method, JHEP 10 (2015) 052, [1503.06849].DOI
  6. [6]P. Nason and G. P. Salam, Multiplicative-accumulative matching of NLO calculations with parton showers, JHEP 01 (2022) 067, [2111.03553].DOI
  7. [7]P. Sarmah, A. Siód­mok and J. Whitehead, KrkNLO matching for colour-singlet processes, JHEP 01 (2025) 062, [2409.16417].DOI
  8. [8]M. van Beekveld, S. Ferrario Ravasio, J. Helliwell, A. Karlberg, G. P. Salam, L. Scyboz et al., Logarithmically-accurate and positive-definite NLO shower matching, JHEP 10 (2025) 038, [2504.05377].DOI
  9. [9]P. Sarmah, A. Siód­mok and J. Whitehead, KrkNLO matching and phenomenology for vector boson processes, 2511.16605.arxiv.org/abs/2511.16605
  10. [10]Y. Chen and R. Frederix, MacNLOPS for ZZ Pair Production at the LHC, Eur. Phys. J. C 86 (2026) 870, [2605.22108].
  11. [11]R. Frederix, S. Frixione, S. Prestel and P. Torrielli, On the reduction of negative weights in MC@NLO-type matching procedures, JHEP 07 (2020) 238, [2002.12716].DOI
  12. [12]K. Danziger, S. Höhe and F. Siegert, Reducing negative weights in Monte Carlo event generation with Sherpa, 2110.15211.arxiv.org/abs/2110.15211
  13. [13]R. Frederix and P. Torrielli, A new way of reducing negative weights in MC@NLO, Eur. Phys. J. C 83 (2023) 1051, [2310.04160].arxiv.org/abs/2310.04160
  14. [14]J. Whitehead, Monte Carlo efficiency via negative weight reduction, EPJ Web Conf. 337 (2025) 01322.
  15. [15]P. Shyamsundar, ARCANE Reweighting: A Monte Carlo Technique to Tackle the Negative Weights Problem in Collider Event Generation, 2502.08052.arxiv.org/abs/2502.08052
  16. [16]P. Shyamsundar, A Demonstration of ARCANE Reweighting: Reducing the Sign Problem in the MC@NLO Generation of e+e− → qq̄ + 1 jet Events, 2502.08053.arxiv.org/abs/2502.08053
  17. [17]S. El Farkh, R. Frederix and M. Gouighri, MC@NLO event generation by reweighting unweighted Born events, Eur. Phys. J. C 86 (2026) 751, [2602.18124].
  18. [18]A. Valassi, MadVfold: accelerating NLO event generation and reducing negative weights with SIMD vectorization and GPUs, 2609.23500.arxiv.org/abs/2609.23500
  19. [19]J. R. Andersen, C. Gütschow, A. Maier and S. Prestel, A Positive Resampler for Monte Carlo Events with Negative Weights, Eur. Phys. J. C 80 (2020) 1007, [2005.09375].
  20. [20]B. Nachman and J. Thaler, Neural resampler for Monte Carlo reweighting with preserved uncertainties, Phys. Rev. D 102 (2020) 076004, [2007.11586].DOI
  21. [21]J. R. Andersen and A. Maier, Unbiased elimination of negative weights in Monte Carlo samples, Eur. Phys. J. C 82 (2022) 433, [2109.07851].DOI
  22. [22]J. R. Andersen, A. Maier and D. Maître, Efficient negative-weight elimination in large high-multiplicity Monte Carlo event samples, Eur. Phys. J. C 83 (2023) 835, [2303.15246].DOI
  23. [23]J. R. Andersen, A. Cueto, S. P. Jones and A. Maier, A Cell Resampler study of Negative Weights in Multi-jet Merged Samples, 2411.11651.arxiv.org/abs/2411.11651
  24. [24]B. Nachman and D. Noll, Neural refinement of sample weights, Phys. Rev. D 112 (2025) 096009, [2505.03724].DOI
  25. [25]C. Palmer and B. Kronheim, Improving statistical precision in Monte Carlo samples with negative weights via reweighting and uncertainty quantification, Phys. Rev. D 113 (2026) 012003, [2510.16217].DOI
  26. [26]J. R. Andersen, E. Cole and A. Maier, Precision Cell Resampling with a Relative and Resonant Aware Metric, 2605.13237.arxiv.org/abs/2605.13237
  27. [27]R. Doherty, L. Hay, R. Jain, M. LeBlanc, J. Marrinan, C. Mauceri et al., Optimal-Transport-Based Cell Resampling for Negative and Pathological Event Weights, 2607.08723.arxiv.org/abs/2607.08723
  28. [28]L. Hay, R. Jain, M. LeBlanc and J. Roloff, Optimizing Cell-Based Negative Weight Mitigation with Optimal Transport, 2609.30357.arxiv.org/abs/2609.30357
  29. [29]A. Butter, T. Plehn and R. Winterhalder, How to GAN event subtraction, SciPost Phys. Core 3 (2020) 009, [1912.08824].DOI
  30. [30]B. Stienen and R. Verheyen, Phase space sampling and inference from weighted events with autoregressive flows, SciPost Phys. 10 (2021) 038, [2011.13445].DOI
  31. [31]L. Breiman, J. Friedman, R. Olshen and C. Stone, Classification and Regression Trees. Wadsworth and Brooks, Monterey, CA, 1984.DOI
  32. [32]J. R. Quinlan, Induction of decision trees, Machine Learning 1 (1986) 81–106.DOI
  33. [33]J. R. Quinlan, C4.5: Programs for Machine Learning. Morgan Kaufmann Publishers, San Mateo, CA, 1993.
  34. [34]Scikit-learn developers, “Decision Trees.” User Guide, official documentation.
  35. [35]Scikit-learn developers, “DecisionTreeRegressor.” Official API documentation.
  36. [36]Scikit-learn developers, “Ensembles: Gradient boosting, random forests, bagging, voting, stacking.” User Guide, official documentation, section on gradient tree boosting.
  37. [37]E. G. Tabak and C. V. Turner, A Family of Nonparametric Density Estimation Algorithms, Commun. Pure Appl. Math. 66 (2013) 145–164.DOI
  38. [38]O. Rippel and R. P. Adams, High-dimensional probability estimation with deep density models, 2013.arxiv.org/abs/1302.5125
  39. [39]L. Dinh, D. Krueger and Y. Bengio, NICE: Non-linear independent components estimation, arXiv preprint arXiv:1410.8516 (2014).DOI
  40. [40]D. Rezende and S. Mohamed, Variational inference with normalizing flows, in International Conference on Machine Learning, pp. 1530–1538, PMLR, 2015.arxiv.org/abs/1505.05770
  41. [41]G. Papamakarios, E. Nalisnick, D. Jimenez Rezende, S. Mohamed and B. Lakshminarayanan, Normalizing Flows for Probabilistic Modeling and Inference, J. Mach. Learn. Res. 22(57) (2021) 1–64 JMLR 19-1028.
  42. [42]J. Alwall, R. Frederix, S. Frixione, V. Hirschi, F. Maltoni, O. Mattelaer et al., The automated computation of tree-level and next-to-leading order differential cross sections, and their matching to parton shower simulations, JHEP 07 (2014) 079, [1405.0301].DOI
  43. [43]S. Frixione, Z. Kunszt and A. Signer, Three jet cross-sections to next-to-leading order, Nucl. Phys. B 467 (1996) 399–442, [hep-ph/9512328].DOI
  44. [44]S. Frixione, A General approach to jet cross-sections in QCD, Nucl. Phys. B 507 (1997) 295–314, [hep-ph/9706545].DOI
  45. [45]R. Frederix, S. Frixione, F. Maltoni and T. Stelzer, Automation of next-to-leading order computations in QCD: The FKS subtraction, JHEP 10 (2009) 003, [0908.4272].DOI
  46. [46]R. D. Ball et al., Parton distributions with LHC data, Nucl. Phys. B 867 (2013) 244–289, [1207.1303].DOI
  47. [47]T. Sjöstrand, S. Ask, J. R. Christiansen, R. Corke, N. Desai, P. Ilten et al., An introduction to PYTHIA 8.2, Comput. Phys. Commun. 191 (2015) 159–177, [1410.3012].DOI
  48. [48]C. Bierlich, S. Chakraborty, N. Desai, L. Gellersen, I. Helenius, P. Ilten et al., A comprehensive guide to the physics and usage of PYTHIA 8.3, SciPost Phys. Codebases (2022) 8, [2203.11601].arxiv.org/abs/2203.11601
  49. [49]R. Frederix and P. Torrielli, Born spreading with consistent shower starting scales in MC@NLO and FxFx merging, 2611.06769.
  50. [50]T. Heimel, N. Huetsch, F. Maltoni, O. Mattelaer, T. Plehn and R. Winterhalder, The MadNIS reloaded, SciPost Phys. 17 (2024) 023, [2311.01548].DOI
  51. [51]C. Durkan, A. Bekasov, I. Murray and G. Papamakarios, Neural Spline Flows, in Advances in Neural Information Processing Systems 32, 6, 2019. 1906.04032.arxiv.org/abs/1906.04032
  52. [52]J. Köhler, A. Krämer and F. Noé, Smooth Normalizing Flows, in Advances in Neural Information Processing Systems 34, 2021. 2110.00351.arxiv.org/abs/2110.00351

Paper details

vlnc:W_ve9pggq9
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.

  • Physics and Astronomy

ChatGPT, Claude

View statement
Looking for related papers…
Contents