Accelerated cross-section predictions with XSEC

Jeriek Van den Abeele

Tools 2020

@JeriekVda

Based on arXiv:2006.16273 with A. Buckley, A. Kvellestad, A. Raklev, P. Scott, J. V. Sparre, and I. A. Vazquez-Holm

IP2I, Lyon

Global fits address the need for a consistent comparison of BSM theories to all relevant experimental data

Challenge:

Scanning increasingly high-dimensional parameter spaces with varying phenomenology

Exploration of a combined likelihood function:

\(\mathcal{L} = \mathcal{L}_\mathsf{collider} \times  \mathcal{L}_\mathsf{Higgs} \times  \mathcal{L}_\mathsf{DM} \times  \mathcal{L}_\mathsf{EWPO} \times  \mathcal{L}_\mathsf{flavour} \times \ldots\)

See GAMBIT talk Friday!

Global fits address the need for a consistent comparison of BSM theories to all relevant experimental data

Challenge:

Scanning increasingly high-dimensional parameter spaces with varying phenomenology

Exploration of a combined likelihood function:

\(\mathcal{L} = \mathcal{L}_\mathsf{collider} \times  \mathcal{L}_\mathsf{Higgs} \times  \mathcal{L}_\mathsf{DM} \times  \mathcal{L}_\mathsf{EWPO} \times  \mathcal{L}_\mathsf{flavour} \times \ldots\)

Global fits need quick, but sufficiently accurate theory predictions

BSM scans today easily require                     \(\sim 10^7\) samples or more

 

Higher-order BSM production cross-sections and theoretical uncertainties make a significant difference!

[GAMBIT, 1705.07919]

CMSSM

[hep-ph/9610490]

Existing higher-order evaluation tools are insufficient for large MSSM scans

  • Prospino/MadGraph: full calculation, minutes/hours per point
  • N(N)LL-fast: fast grid interpolation, but only degenerate squark masses

$$ pp\to\tilde g \tilde g,\ \tilde g \tilde q_i,\ \tilde q_i \tilde q_j, $$

$$\tilde q_i \tilde q_j^{*},\ \tilde b_i \tilde b_i^{*},\ \tilde t_i \tilde t_i^{*}$$

at \(\mathsf{\sqrt{s}=13} \) TeV

xsec 1.0 performs smart regression for strong SUSY processes at the LHC

  • New Python tool for NLO cross-section predictions within seconds

A. Buckley, A. Kvellestad, A. Raklev, P. Scott, J. V. Sparre, JVDA, I. A. Vazquez-Holm

  • Pointwise estimates of PDF, \(\alpha_s\) and scale uncertainties, in addition to the subdominant regression uncertainty
  • Future expansions: other processes, more CoM energies, higher-order corrections, ...

$$$

A quick sketch

Interpolation ...

Interpolation ... doesn't give prediction uncertainty!

Correlation length-scale

Correlation length-scale

Estimate from data!

Correlation length-scale

Estimate from data!

Correlation length-scale

Gaussian process prediction with uncertainty

Estimate from data!

prior distribution over all functions

with the estimated smoothness

f(\vec x) \sim \mathcal{GP}\left[m(\vec x), k(\vec x, \vec x')\right]
\mathrm{E}\left[f(\vec x)\right] = m(\vec x)
\mathrm{Cov}\left[f(\vec x), f(\vec x')\right] = k(\vec x, \vec x')

The Bayesian Way: quantify beliefs with probability

k(\vec x, \vec x') = \sigma_f^2 \exp\left(-\frac{|\vec x-\vec x'|^2}{l^2}\right)

prior distribution over all functions

with the estimated smoothness

f(\vec x) \sim \mathcal{GP}\left[m(\vec x), k(\vec x, \vec x')\right]
\mathrm{E}\left[f(\vec x)\right] = m(\vec x)
\mathrm{Cov}\left[f(\vec x), f(\vec x')\right] = k(\vec x, \vec x')

posterior distribution over functions

with updated \(m(\vec x)\)

data

The Bayesian Way: quantify beliefs with probability

k(\vec x, \vec x') = \sigma_f^2 \exp\left(-\frac{|\vec x-\vec x'|^2}{l^2}\right)

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

Let's make some draws from this prior distribution

At each input point, we obtain a distribution of possible function values (prior to looking at data)

At each input point, we obtain a distribution of possible function values (prior to looking at data)

At each input point, we obtain a distribution of possible function values (prior to looking at data)

A Gaussian process sets up an infinite number of correlated Gaussians, one at each parameter point

We obtain a posterior by conditioning on known data

target function

We obtain a posterior by conditioning on known data

target function

We obtain a posterior by conditioning on known data

We obtain a posterior by conditioning on known data

Make posterior predictions at new points by computing correlations to known points

Make posterior predictions at new points by computing correlations to known points

Posterior predictive distributions are Gaussian too

Squared Exponential kernel

Matérn-3/2 kernel

Different kernels lead to different function spaces to marginalise over

Workflow

Generating data

Random sampling

SUSY spectrum

Cross-sections

Optimise kernel hyperparameters

Training GPs

GP predictions

Input parameters

Linear algebra

Cross-section

estimates

Compute covariances between training points

Workflow

Generating data

Random sampling

SUSY spectrum

Cross-sections

Optimise kernel hyperparameters

Training GPs

GP predictions

Input parameters

Linear algebra

Cross-section

estimates

XSEC

Compute covariances between training points

Training scales as \(\mathcal{O}(n^3)\), prediction as \(\mathcal{O}(n^2)\)

A balancing act

Random sampling with different priors, directly in mass space

Evaluation speed

Sample coverage

Need to cover a large parameter space

Distributed Gaussian processes

[Liu+, 1806.00720]

Distributed Gaussian Processes

Divide-and-conquer approach for dealing with large datasets

  • Make local predictions with smaller data subsets
  • Compute a weighted average of these predictions

The exact weighting procedure is important, to ensure

  • Smooth final predictions
  • Valid regression uncertainties

"Generalized Robust Bayesian Committee Machine"

Validation results

Gluino-gluino

Gluino-squark

Squark-squark

Squark-antisquark

Fast estimate of SUSY (strong) production cross- sections at NLO, and uncertainties from

  • regression itself
  • renormalisation scale
  • PDF variation
  • \(\alpha_s\) variation

Goal

$$ pp\to\tilde g \tilde g,\ \tilde g \tilde q_i,\ \tilde q_i \tilde q_j, $$

$$\tilde q_i \tilde q_j^{*},\ \tilde b_i \tilde b_i^{*},\ \tilde t_i \tilde t_i^{*}$$

Interface

Method

Pre-trained, distributed Gaussian processes

Python tool with command-line interface

Processes

at \(\mathsf{\sqrt{s}=13}\) TeV

Tutorial on Friday!

Thank you!

Backup slides

Some linear algebra

Regression problem, with 'measurement' noise:

\(y=f(\vec x) + \varepsilon, \ \varepsilon\sim \mathcal{N}(0,\sigma_n^2) \quad \rightarrow \quad \) infer \(f\), given data  \(\mathcal{D} = \{\vec X, \vec y\}\)

Assume covariance structure expressed by a kernel function, like

k(\vec x_i, \vec x_j) = \sigma_f^2\, \exp\!\left(-\frac{|\vec x_i - \vec x_j|^2}{2\mathcal{l}^2}\right) + \sigma_n^2\, \delta_{ij}

Consider the data as a sample from a multivariate Gaussian distribution

\([\vec x_1, \vec x_2, \ldots]\)

\([y_1, y_2, \ldots]\)

signal kernel

white-noise kernel

Some linear algebra

Regression problem, with 'measurement' noise:

\(y=f(\vec x) + \varepsilon, \ \varepsilon\sim \mathcal{N}(0,\sigma_n^2) \quad \rightarrow \quad \) infer \(f\), given data  \(\mathcal{D} = \{\vec X, \vec y\}\)

Training: optimise kernel hyperparameters by maximising the marginal likelihood

p\left( \vec y\ |\ \vec X, \vec \theta \right) = \mathcal{N} \left( \vec y \ |\ m(\vec X), K_{\vec\theta} \right)
y_*\, |\, \mathcal D, \vec x_* \sim \mathcal{N}\left(\mu_*, \sigma_*^2\right)
\propto |K_{\vec\theta}|^{-\frac{1}{2}} \times \exp \left\{ -\frac{1}{2} \left( \vec y - m(\vec X) \right)^T K_{\vec\theta}^{-1} \left(\vec y - m(\vec X) \right) \right\} % %%k(\vec X, \vec X)^{-1} \left( \vec y - \mu(\vec X) \right), % k(\vec x_*, \vec x_*) - k(\vec x_*, \vec X) k(\vec X, \vec X)^{-1} k(\vec X, \vec x_*)

Posterior predictive distribution at a new point  \(\vec x_*\) :

with

\mu_* = m(\vec x_*) + K_* K^{-1}(\vec{y}-m(\vec X)) %%\tilde\mu(\vec x_*) = \mu(\vec x_*) + k(\vec x_*, \vec X) k(\vec X, \vec X)^{-1}\vec{y}\\ %\tilde\sigma^2(\vec x_*) = k(\vec x_*, \vec x_*) - k(\vec x_*, \vec X) k(\vec X, \vec X)^{-1} k(\vec X, \vec x_*)

Implicit integration over points not in  \(\vec X\)

\sigma_*^2 = K_{**}-K_*K^{-1}K_*^T %%\tilde\mu(\vec x_*) = \mu(\vec x_*) + k(\vec x_*, \vec X) k(\vec X, \vec X)^{-1}\vec{y}\\ %\tilde\sigma^2(\vec x_*) = k(\vec x_*, \vec x_*) - k(\vec x_*, \vec X) k(\vec X, \vec X)^{-1} k(\vec X, \vec x_*)

[

K = k(\vec X, \vec X) \\ K_* = k(x_*, \vec X) \\ K_{**} = k(\vec x_*, \vec x_*)

Scale-dependence of LO/NLO

[Beenakker+, hep-ph/9610490]

Made with Slides.com