Model-Independent Bounds on a Light Higgs

TL;DR

This work develops a model-independent framework to bound Higgs-like scalars by recasting LHC search results into a generic coupling space (a,c) via a reconstruction of channel likelihoods from publicly available 95% CL limits. It formulates an electroweak chiral Lagrangian with a light Higgs and shows how composite-Higgs scenarios modify couplings, then introduces a Gaussian-likelihood method to combine channels and extract constraints without full experimental likelihoods. Applying this to current data, the authors derive bounds on a and c, map ξ in MCHM4/5, and analyze the 125 GeV excess, finding multiple viable regions depending on channel and production mode. The study demonstrates the value of exclusive analyses and calls for publication of detailed channel information to enable precise, model-independent Higgs coupling inferences.

Abstract

We present up-to-date constraints on a generic Higgs parameter space. An accurate assessment of these exclusions must take into account statistical, and potentially signal, fluctuations in the data currently taken at the LHC. For this, we have constructed a straightforward statistical method for making full use of the data that is publicly available. We show that, using the expected and observed exclusions which are quoted for each search channel, we can fully reconstruct likelihood profiles under very reasonable and simple assumptions. Even working with this somewhat limited information, we show that our method is sufficiently accurate to warrant its study and advocate its use over more naive prescriptions. Using this method, we can begin to narrow in on the remaining viable parameter space for a Higgs-like scalar state, and to ascertain the nature of any hints of new physics---Higgs or otherwise---appearing in the data.

Figures (10)

• Figure 1: Limits on the coupling $a^2$ implied by the LEP precision tests for $\Lambda = 4\pi v/\sqrt{1-a^2}$ and $m_t =173.2\,$GeV. The gray region is excluded at 99% CL.
• Figure 2: Posterior probability $p(\mu|n_{obs})$ obtained for $n_{obs} = 35$, $n_b=30$, $n_s^{SM}=3$ (continuous curve). In this example the maximum is at $\mu_{max} = 5/3$, and the 95% CL limit on $\mu$ is $\mu^{95\%}_{obs} = 5.66$. The dashed curve shows the approximating Gaussian with mean $\mu_{max}$ and standard deviation $\sigma_{obs} = \sqrt{35}/3$.
• Figure 3: Left panel: 95% CL observed limits on $\mu$ obtained by combining all CMS searches with different techniques: the continuous black curve is the official CMS limit, the dotted red and dashed orange curves are obtained respectively with our method and by a naive quadrature combination. Right panel: relative deviation of the limits obtained with these two latter approaches from the official combination. The blue band at $\pm 20\%$ is for illustration.
• Figure 4: Relative error between extracted and constructed likelihoods for the five $h \to WW$ categories of CMS, as a function of the signal strength modifier $\mu$. In each case the extracted Gaussian likelihood is found to approximate the one constructed from event numbers typically to within $20\%$.
• Figure 5: 95% CL observed limits in the plane $(a,c)$ obtained by combining the five $WW$ categories in CMS for $m_h = 120\,$GeV. The blue and orange curves are obtained using respectively the likelihoods constructed from the number of events in Table \ref{['table:WW120']} (exact combination) and the likelihoods reconstructed with our method (gaussian approximation).