Testing naturalness at 100 TeV

Solutions to the electroweak hierarchy problem typically introduce a new symmetry to stabilize the quadratic ultraviolet sensitivity in the self-energy of the Higgs boson. The new symmetry is either broken softly or collectively, as for example in supersymmetric and little Higgs theories. At low energies such theories contain naturalness partners of the Standard Model fields which are responsible for canceling the quadratic divergence in the squared Higgs mass. Post the discovery of any partner-like particles, we propose to test the aforementioned cancellation by measuring relevant Higgs couplings. Using the fermionic top partners in little Higgs theories as an illustration, we construct a simplified model for naturalness and initiate a study on testing naturalness. After electroweak symmetry breaking, naturalness in the top sector requires aT = − λt2 at leading order, where λt and aT are the Higgs couplings to a pair of top quarks and top partners, respectively. Using a multivariate method of Boosted Decision Tree to tag boosted particles in the Standard Model, we show that, with a luminosity of 30 ab−1 at a 100 TeV pp-collider, naturalness could be tested with a precision of 10% for a top partner mass up to 2.5 TeV.


Introduction
The naturalness problem [1], motivated by the smallness of the Standard Model (SM) particle masses in comparison to a UV cutoff, has been the driving force behind theoretical developments in particle physics for several decades. Especially after the discovery of the 125 GeV Higgs boson at the Large Hadron Collider (LHC) in 2012 [2,3], there is a renewed urgency to understand why its mass is so much smaller than the next known scale in particle physics, the Planck scale, given the quadratic ultraviolet sensitivity in the selfenergy of the Higgs. In the past decades, many theories of naturalness have been proposed. Typically a new symmetry which protects the squared mass of the SM-like Higgs boson from quadratic UV-sensitivity is introduced. Two eminent examples are supersymmetry [4] and spontaneously broken global symmetry [5,6], which require the existence of naturalness partners coupling to the Higgs boson. The symmetry-protected relation among the Higgs couplings with the SM particles and their partners, called the naturalness sum rule, ensures cancellation of the quadratic divergence in the squared Higgs mass. Confirming or refuting theories of naturalness is one of the most important goals for the current and future high energy collider experiments.
Among the naturalness partners, the partner of the SM top quark typically plays the most prominent role. Currently, a tremendous amount of effort at the LHC has been going into searches for top partners [7][8][9][10], whose discovery potential at a next-generation hadron collider has also been studied [11]. However, the discovery of a partner-like particle is only the first step in confirming theories of naturalness. In order to ensure that the discovered partner-like particle is not a naturalness impostor, but indeed the one that yields the cancellation of the quadratic UV-sensitivity in the Higgs self-energy, one needs to further measure the naturalness sum rule. In this article, we propose a novel way to JHEP09(2017)129 probe the naturalness sum rule at hadron colliders, in order to facilitate the testability of theories of naturalness. For convenience, we rephrase the naturalness sum rule ∆m 2 H SM + ∆m 2 H NP = 0, which relates the different contributions to the quadratically divergent quantum corrections to the squared Higgs mass ∆m 2 H , into a naturalness parameter Note that in the SM the top quarks contribute the strongest to the UV-sensitivity of the Higgs self-energy. As a first illustration for testing naturalness, we will analyze fermionic top partners which carry QCD colors, as predicted by little Higgs models [12][13][14]. 1 Similar colored top partners exist in the so-called composite Higgs models as well [15,16]. We define a simplified model consisting of the SM top fields and a pair of vector-like fermionic top-partners, which are either singlets or doublets under SU(2) w . Then we show that the naturalness sum rule leads to an especially simple form after electroweak symmetry breaking, at leading order in the Higgs vacuum expectation value (VEV) here m T is the top partner mass and the relation involves at leading order only the couplings of the Higgs to a pair of top quarks λ t and a pair of top-partners a T . This suggests that, in order to test the naturalness sum rule up to O v 2 /m 2 T , only two measurements are necessary, one for each of the couplings λ t and a T . We then proceed to study how this relation can be measured to quantify the degree of cancellation in the contribution of the top sector to the quadratic divergence of the Higgs self energy in a future 100 TeV hadron collider [17,18].
In section 2 we introduce a simplified model for naturalness, by extending the SM top sector with a pair of vector-like top partners, and derive relation from the requirement of naturalness. Based on this discussion we design a detector study dedicated to test naturalness of fermionic top partners at a 100 TeV collider in section 3. Finally we conclude in section 4.

A simplified model for naturalness
In the SM, the UV-sensitivity of the Higgs self-energy originates from three major sources which, in order of decreasing magnitude, are the coupling to top quarks, electroweak gauge bosons and the Higgs boson itself. In a typical natural theory, the contributions of top quarks, EW gauge bosons and the Higgs boson are expected to be canceled by their respective partner particles. In this work we consider a top sector with fermionic partner particles, which carry QCD colors and could be produced copiously in a hadron collider, if their masses are not too heavy. These top partners are vector-like and massive, thus allowing a quartic coupling with H † H which enables the cancellation of the top quadratic divergence in the Higgs self-energy. We focus on the cases where the vector-like top partners carry the same EW charges as either the right-chiral top quark u c 3 [14,19] The mass dimension in these couplings is given by the symmetry breaking scale f and we imply |H| 2 = H † H. The parameter c i and c i are free and for odd-dimensional couplings they can be complex. A linear combination of u c 3 and U c together with U becomes massive with a top partner mass of Whereas, the orthogonal combination remains massless and acquires a Yukawa coupling to the Higgs doublet. In terms of these mass eigenstates where the couplings are given by The case in which the vector-like pair (Q c , Q) consists of SU(2) w doublets can be written as The rotation into mass eigenstates leads to a Lagrangian identical to (2.4) with the replacement of

JHEP09(2017)129
Although the following discussion is based on Lagrangian (2.1) it can also be applied to Lagrangian (2.6) and our results can easily be adjusted to correspond to that case. The contribution of Lagrangian (2.4) to the Higgs potential can be computed using the Coleman-Weinberg potential [21] and the quadratically divergent part is where M(H) is the fermion mass matrix with the Higgs field H as a background field, and Λ is the cutoff scale of the theory. In the basis (t c , T c ) and (t , T ) the mass matrix reads up to quadratic order in the Higgs field The absence of quadratically divergent contributions in the Higgs mass is then equivalent to the requirement that the coefficient of the |H| 2 term in tr M 2 vanishes Several comments follow from this equation. First, due to the structure of the matrix (2.9), no linear term and hence no tadpole contribution arises. Second, α t only contributes to |H| 4 terms and does not enter into the cancellation of quadratic divergences in the Higgs mass. Finally, requiring the coefficients of the |H| 2 term to vanish leads to the naturalness sum rule It is worthwhile pointing out that the λ T term contributes to the quadratic divergences with the same sign as the SM top quark, and the cancellation hinges entirely on the existence of the four-point coupling α T which must have a negative sign. In terms of the parameter in the Lagrangians of the gauge eigenstates (2.1) and (2.6) this naturalness sum rule reads The best known examples of models described by Lagrangian (2.4) and subject to the naturalness sum rule (2.12) are little Higgs models. Some typical examples are presented in table 1. 2 The collective symmetry breaking in the Yukawa sector of little Higgs models yields the relations Mirror twin Higgs SU(4) U(1) as can easily be checked in table 1. This ensures at one-loop level the cancellation of the quadratic divergences from the top sector according to the condition (2.13). A common but by no means necessary additional restriction is to couple only one of the top partners to the Higgs field by setting c 1 = c 2 = · · · = 0, which leads to For phenomenological reasons it can be desirable to introduce T -parity [25,26]. When T -parity is implemented in the top sector [19] the top partner particles are T -odd, while SM particles are T -even. Therefore the mixing terms have to vanish β T = α t = λ T = 0. Before rotation into the mass eigenstates this corresponds to the condition c i = ∓c i for even and odd i, respectively.
Finally, expanding the Higgs field around its VEV 3 and rotating the fermion fields to linear order in v into mass eigenstates

JHEP09(2017)129
leads to the Lagrangian The couplings m T , λ t,T and α t,T are up to linear order in v /m T identical to their primed counterparts, and we have collected the higher order correction in appendix A. The new VEV induced couplings are The naturalness sum rule (2.12) leads at leading order in v to the remarkable relation (2.20) Though the naturalness cancellation in Little Higgs models applies at one-loop only, the high-loop corrections to the sum rule are suppressed by extra loop factors. They are negligibly small compared to the uncertainty in this relation unless m T O(1) TeV. Additionally, in the simplified model under consideration higher dimensional operators encoding new physics at the cutoff scale Λ ∼ 4πf lead to terms of higher order in Lagrangian (2.1) and (2.6). Our calculations show that these terms do not influence the relation in (2.20).
The naturalness sum rule (2.20) has significant implications on tests of the naturalness sum rule at colliders. Naively, the relation (2.12) suggests that three measurements, one for each of the couplings λ t , λ T and α T , are necessary in order to test naturalness. However, as λ t , λ T and α T are not defined in the basis of mass eigenstates after electroweak symmetry breaking, the separate reconstruction of their values becomes rather challenging. Previously it was suggested [22] to test the naturalness relation (2.12) by measuring the λ T coupling via the top partner decay T → th, and extracting the information of α T with the relation However, the proposed strategy is limited by two observations. First this relation is not generic for all little Higgs models. The λ T term for example can be eliminated by imposing T -parity in the top sector [19]. Second, it implies that in order to extracting the value of the coupling α T it is necessary to measure the decay constant f . In contrast, the naturalness sum rule (2.20) has the virtue of being generic for all models which can be described by the simplified model under consideration. Especially it is independent of the assumption on T -parity. It is exact up to order v /m T , and involves only two free parameters, the coupling of the Higgs to two top quarks and to two top partners. Notably, this does not involve the measurement of the decay constant f . In conclusion the relation (2.20) greatly simplifies testing the naturalness sum rule.

JHEP09(2017)129
Although, the naturalness sum rules (2.12), (2.13) and (2.20) are derived for vectorlike top partners in little Higgs models, they can also be applied to mirror twin Higgs models [24]. In this case, top quarks carry QCD charge while top partners carry mirror QCD charge, which forbids mixing between them. This is ensured by the condition that c 0 = c 1 = c 2 ≡ 0, which yields the relations β T = α t = λ T = 0. Note the similarity to the case of little Higgs models with T -parity. However, in this case the naturalness sum rule (2.13) is reduced to Although, the collider kinematics of colorless top partners differs significantly from that of colored top partners, we list twin Higgs models as one possibility in table 1. However, the collider analysis in the following section applies only to models with colored top partners such as little Higgs models.

Collider analysis
In this section we assume that one of the searches for fermionic top partners [7][8][9][10][11] has been successful and that the mass m T of the newfound particle has been determined. Based on the discussions above, the measurements of the couplings of a Higgs with a pair of top quarks and a pair of top partners constitute the two cornerstones for testing the naturalness sum rule. The former measurement has been extensively discussed in the literature, for searches at the LHC [27][28][29][30][31][32], at a future e + e − collider [33][34][35], and at a 100 TeV hadron collider [36]. We will focus on the measurement of the Higgs coupling to two top partners a T via T c T h production at a 100 TeV hadron collider. The naturalness parameter (1.1) applied to this analysis reads If the naturalness sum rule (2.12) is satisfied µ t is up to an order of v 2 /m 2 T equal to one. In the following we assume the particle content and interactions of the SM augmented by a colored fermionic top partner subject to Lagrangian (2.18) without T -parity. The analysis proposed here is blind to the sign of a T and we leave the removal of this degeneracy for a future publication.

Simulation of signal and background
In little Higgs models, fermionic top partners decay mainly into heavy quarks and bosons B = W ± , Z, h. The equivalence theorem ensures that in the limit f v the branching ratios (BR) relate according to [37] BR(T → th) BR(T → tZ) This relation receives a correction of order v /m T from the a t term in   For simplicity we ignore this uncertainty in the current analysis by fixing the BRs to the values in relation (3.2).
In order to be more consistent with the majority of the literature on collider phenomenology we will switch the notation from using Wely fermions (T c , T ) to Dirac fermions T , T for the remaining part of this section. We abbreviate heavy neutral bosons by B 0 = Z, h and assume that jets j also include b-jets. The decay channels in this notation are Here we assume that the associated Higgs is unboosted and decays into two separate bottom jets, while the other bosons are boosted and decay hadronically within the cone radius of a single jet. Furthermore, we require the top quarks to decay either fully or semi-leptonically. We have implemented Lagrangian (2.18) together with the coupling which ensure the BRs (3.2) in FeynRules 2.3 [38]. This enables us to generate events with MadGraph 2.4 [39] and decay heavy particles with MadSpin. Light particle decay and showering is done in Pythia 6.4 [40] followed by the detector simulation Delphes 3.3 [41].
We have calculated the LO production cross section of the T T h process with MadGraph, and present it for 14 TeV and 100 TeV as a function of the top partner mass m T and the naturalness parameter µ t in figure 1. In particular, we neglect the Higgs coupling to two gluons, which is induced at one-loop level and formally a next-to-leading order contribution. However, when |µ t | is small, the tree-level LO production rate becomes small and the oneloop contribution arising from the Higgs coupling to two gluons could become dominant. For this reason we choose to focus on the region of parameter space where |µ t | > 0.4. We have checked that in this case the tree-level production rate still dominates. For simplicity

Cross section [fb]
semi-leptonic full-leptonic p T 3150 526 n(l isol ) 1315 111 Table 2. Cross section of the background processttjjjj at 100 TeV in fb. During event generation we require at least two quarks to be bottom or charm quarks with p T > 40 GeV. The other two quarks are required to be boosted with p T > 400 GeV. The resulting cross sections are given in the row labeled p T . Before the analysis we require the presence of at least one and two isolated leptons for semi-and full hadronic events, respectively. The isolation criterion is given at the end of section 3.1. We give the cross section after this pre-cut in the row labeled n(l isol ).
we have also assumed the top Yukawa coupling λ t takes the value known from the SM, up to corrections of order v 2 /m 2 T . In order to reduce the hadronic background we focus on the decay channel (3.4a) which involves two top quarks. The dominant electroweak background for this signal consists of a pair of top quarks produced in association with three bosons. In this case each top partner is mimicked by the combination of one top quark with one electroweak boson. However, the top pair production in association with multiple quarks has a much larger cross section. If we consider the mis-identification of jets we have to distinguish the cases in which pairs of jets mimic reconstructed objects from the cases in which a singe jet is mistaken for a single boosted object. On the one hand, the decay products of heavy top-partner are generically boosted and tend to be detected as a single jet. On the other hand, the Higgs is produced with a low transverse momentum and its decay products will be detected separately. In this study we consider only Higgs decays into bottom quarks. In order to reduce the size of the top pair plus jet background we require the two leading jets to be boosted enough to lie within a jet cone of ∆R ≈ 0.5 We expect these jets to mimic the bosons originating from the top partner decay. Additionally, a combination of two softer jets can be misidentified as being the Higgs boson. As we require the Higgs boson signature to be b-like we restrict these jets to come from c or b quarks. Therefore, the main background in this analysis is top pair production in association with four jets where the two leading once are boosted while the other two are b-like. We have generated this background with MadGraph and have required the jet transverse momentum to be p T > 400 and 40 GeV for the two leading and the remaining jets, respectively. We show the LO cross sections for semi-and full leptonic decays of the top quarks in table 2. As a cut on the jet number is only a weak discriminator the top-pair production with more than four jets will contribute to the background. Because the strong coupling constant appears in an higher power and due to the increased number of jets which leads to a larger phase space this higher order background will be suppressed with respect to the leading contribution. We do not consider any k-factor in this analysis.

JHEP09(2017)129
Signal acceptance  (We are using an effective model in order to produce Higgs boson pairs.) During production and tagging we have ensured that the transverse momenta are constrained to 400-700 GeV for a and b and to 700-1000 GeV for c and d. As the jet cone size is ∆R = 0.5, we expect the bosons of the SM and the top quarks to lie within a single jet for p T 400 and 700 GeV, respectively.
During the analysis we cluster jets with the anti-k T algorithm using a minimal transverse momentum of 40 GeV and a jet cone size of 0.5. Additionally we require the two leading jets to be boosted more than p T > 400 GeV. We require at least one and two leptons for the semi-and full-leptonic analysis, respectively. Leptons must be isolated with an isolation radius of ∆R > 0.3 and a maximal transverse momentum ratio of 0.2. The transverse momentum ratio is defined between the lepton and other cell activity within the isolation cone. If the lepton transverse momentum is larger than 100 GeV, we do not require it to be isolated [42]. Finally, we disregard events with less than five jets.

Event reconstruction
We reconstruct the signal signatures with multiple boosted decision trees (BDTs). Each of them is optimized to tag or reconstruct a single physical object, such as bottom or top quark, Higgs or Z boson and top partners. Such an object can consist of a sub-jet, a single   jet or be combined from multiple jets. This Boosted Collider Analysis (BoCA) 0.3 [43] has been introduced in [44,45] where we also discuss the efficiency of the bottom and top taggers. BoCA uses the TMVA library [46] of the ROOT 6.06 framework [47] together with FastJet 3.2 [48] for jet clustering. Here we introduce additionally taggers dedicated to Higgs and Z boson tagging. For boosted bosons we demonstrate their background mis-identification as a function of a given signal rate, the so called Receiver Operating Characteristic (ROC) curves in figure 2. We show the results for two different ranges of transverse momentum, which correspond to the cases in which the top quarks are boosted enough to end up in one and two jets, respectively. The Higgs tagger utilizes the displaced vertices of the bottom decay products. The strongest discriminators within these taggers are the boson mass and the pull [49] between its components. Furthermore, we train a neutral boson tagger to detect both of these particles, which allows us to reconstruct top partners from tops and neutral bosons. During this reconstruction we cater towards the boostness of the decay products by using the available substructure information. In order to suppress thett + 3j and thett + 5j backgrounds we require the neutral bosons from the top partner decays to be boosted with a cone size of ∆R < 0.5 while the associated Higgs boson has to be un-boosted with a cone size of ∆R > 0.5. Finally, we reconstruct the T T h signature by combining two top partner tagger with a Higgs boson tagger. The signal is separated from the background with a single cut on the final BDT result.

Results
For n observed events the significance can be defined as the log-likelihood ratio [50] Z(x|n) = −2 ln L(x|n) L(n|n) , where x is the number of events predicted by the hypothesis which is tested. For counting experiments the likelihood function is given by the Poisson probability For the exclusion of the signal hypothesis b+s we require Z(b + s|n) ≥ 2 and for the discovery of a signal by exclusion of the background only hypothesis b we require Z(b|n) ≥ 5. As we are projecting to future experiments we replace the event number n with the prediction for an alternative hypothesis. Hence, we are using Z(b|b + s) ≥ 5 for the discovery of a model and Z(b + s| b + s| nat ) ≥ 2 for the exclusion of an unnatural model. In figure 3, we present the discovery reach against the background-only hypothesis of top partners in the T T h production channel at 100 TeV. This may not be the most sensitive channel for the search of fermionic top partners, but it shows the effectiveness of our analysis for testing naturalness. For natural theories we see that the discovery reach for top partners coupled to a Higgs boson can be pushed up to ∼ 1.4, 2, 2.5 TeV, with 0.3, 3, 30 ab −1 of data, respectively. In comparison, the ATLAS [7][8][9] and CMS [10] experiments have excluded fermionic top partner with m T < 780 GeV, given BRs compatible with (3.2). The systematic errors are not included in our analysis. Instead, we present the expected signal over background ratio in figure 4 which indicates that the contour with s /b = 10 % could extend to above 2.5 TeV for natural theories.
In the following we introduce two measures for the sensitivity of testing the naturalness sum rule. First, we calculate the exclusion limit of unnatural theories with |µ t | = 1, against the assumption that the observation at the collider is consistent with the prediction of a theory with |µ t | = 1. The exclusion limits of unnatural theories are presented in figure 5. With 30 ab −1 of data, unnatural theories with ||µ t | − 1| > 0.1 can be excluded up to m T ∼ 2.2 TeV.
Second, we calculate the precision of measuring the naturalness parameter, defined as the uncertainty at one sigma confidence level (3.8)   The precision of the coupling a T can be estimated with the relation .
In figure 6 we present the precision under the assumption of a perfectly measured top Yukawa coupling δλ t = 0. Figure 6a shows that in natural theories the naturalness parameter could be measured with a precision of 10 % up to a top partner mass of m T ∼ 2 TeV, using 3 ab −1 of data. Figure 6b shows that with 0.   has the consequence that curves for Z(b|n) = 5 presented in figure 3 are approximately identical to the curves for δ|µt| /|µt| = 10 % shown in figure 6.
In figure 6 we have neglected the uncertainty on the measurement of the top Yukawa coupling. However, the measurement of the top Yukawa coupling is non-trivial and will be an important task at a 100 TeV hadron collider [36]. In figure 7, we show the precision of measuring the naturalness parameter when the uncertainty of measuring λ t is included. We consider two cases. In figure 7a we assume a luminosity of 3 ab −1 at 100 TeV and an uncertainty of 10 % achievable in future λ t measurements at the HL-LHC [51,52]. In figure 7b we consider 30 ab −1 and an uncertainty of 1 % potentially reachable at a future 100 TeV collider [36]. Although a precision of 10 % for the measurement of the naturalness parameter could still be pushed up to m T ∼ 2.5 TeV, we do see a significant impact from δλ t , particularly for small m T . Therefore, it is crucial to improve the precision of measuring λ t , in order to test naturalness with a higher precision.

Discussion and conclusion
The naturalness problem plays an extremely important role in driving particle physics. While, searches for partner particles predicted by theories of naturalness are actively conducted at the LHC, a further step is necessary to actually test the naturalness of these particles. Once discovered, it must be ensured that the newly found particle is not an impostor, but a particle with the right properties to cancel the quadratic UV-sensitivity of the Higgs self-energy.
Using the fermionic top sector as an example, we show how the naturalness sum rule can be measured and what sensitivity reach can be achieved at a 100 TeV hadron collider. We developed a simplified model for vector-like fermionic top partners and derived the naturalness sum rule in this general setup. Remarkably, we found that the naturalness sum JHEP09(2017)129 rule only involves couplings of the Higgs to a pair of top quarks and a pair of top partners, and is exact up to an order of v 2 /m 2 T . The derived condition is rather generic, covering representative theories of naturalness, including for example little Higgs models both with and without T -parity and mirror twin Higgs models. In particular, the necessary measurements do not involve the measurement of the decay constant f . The 100 TeV collider analyses based on the T T h production in the little Higgs models indicates that at 30 ab −1 : • Theories which predict a |µ t | deviating from the naturalness assumption by more than 10 % which corresponds to ||µ t | − 1| > 0.1 could be excluded for a top partner mass of up to ∼ 2.2 TeV.
• A precision of 10 % for the measurement of |µ t | could be achieved with a top partner mass of up to ∼ 2.5 TeV, given the precision of measuring the top Yukawa coupling of some percent.
We expect that the reach of this measurement can be further improved. Recall, that in this analysis presented in this article we only consider decays of T T h via a pair of top quarks and require them to decay fully or semi-leptonically. However, a larger branching fraction originates from the decay via one top quark and one bottom quark, that is T T h → tB 0 bW + h or T T h → tB 0b W − h. Additionally, the Higgs boson accompanying the top partners are required to decay to a pair of bottom quarks. In reality, some of its other decay modes may lead to a stronger background suppression. We would expect that these channels bring non-trivial improvements to the sensitivity of testing naturalness.
But, in this analysis we have neglected the precision with which the top partner mass can be measured (e.g., by mass reconstruction in the T T events). This yields an additional uncertainty on the relation between the signal rate and the coupling α T and hence a contribution to δα T . An accurate estimation of the influence of this factor will be necessary. Additionally, we did not take into account systematic errors in the analysis. The systematic errors could receive multiple contributions, e.g., high-order corrections to the QCD background, PDF uncertainty, collider performance of future collider, etc. A reliable estimation is beyond the scope of this article due to the lack of input information. So we would postpone its considerations to future studies. At last, we would like to stress again that the T T h production is insensitive to the sign of the naturalness parameter and hence causes a degeneracy problem, despite its important role in testing naturalness. In order to break this degeneracy and exclude the theories with µ t < 0, new strategies such as indirect detection needs to be introduced.
Given the crucial role of the naturalness problem in driving particle physics, it is worthwhile to extend the analysis pursued in this article to other contexts. On the one hand, a complete cancellation of quadratic UV-sensitivity in the Higgs self-energy in little Higgs models requires the introduction of additional partner particles for gauge and Higgs bosons, which have to fulfill the corresponding naturalness sum rules. On the other hand, other theories of naturalness, most notably supersymmetry, rely on partner particles with opposite statistics in comparison to their corresponding SM particles, yielding a sum rule JHEP09(2017)129 as well. Some relevant discussions can be found in, e.g., ref [53][54][55]. A full exploration of these possibilities will be deferred to future work.