Using Monte Carlo to Estimate Systematic Uncertainties
Advisor(s)
Dr. Peter Dong, Illinois Mathematics and Science Academy
Location
Room A147
Start Date
26-4-2019 10:45 AM
End Date
26-4-2019 11:00 AM
Abstract
In order to interpret data from the CMS particle accelerator, we must compare generator-level variables with reconstructed variables in Monte Carlo samples. These comparisons allow us to estimate systematic uncertainties in the detector.
Generating Monte Carlo samples is a tedious process. Each step (generation, reconstruction, AOD, and miniAOD) entails creating new python files, submitting each file, checking the dashboard for completion, and resubmitting the step if needed. We created a program to automate this process. The program can handle multiple requests for different mass ranges, lambda values, interference types, and helicities. It uses a crontab to check the progress of each step, send emails when each stage is completed, and move to the next step.
Analyzing the error in the detector using Monte Carlo samples will allow us to compensate for the uncertainties when working with real data. We will present studies of invariant mass and transverse momentum resolution and the effect of acceptance and migration as a function of invariant mass. Both studies allow an estimate of key systematic uncertainties in the lepton energy scale that are used in the final analysis.
Using Monte Carlo to Estimate Systematic Uncertainties
Room A147
In order to interpret data from the CMS particle accelerator, we must compare generator-level variables with reconstructed variables in Monte Carlo samples. These comparisons allow us to estimate systematic uncertainties in the detector.
Generating Monte Carlo samples is a tedious process. Each step (generation, reconstruction, AOD, and miniAOD) entails creating new python files, submitting each file, checking the dashboard for completion, and resubmitting the step if needed. We created a program to automate this process. The program can handle multiple requests for different mass ranges, lambda values, interference types, and helicities. It uses a crontab to check the progress of each step, send emails when each stage is completed, and move to the next step.
Analyzing the error in the detector using Monte Carlo samples will allow us to compensate for the uncertainties when working with real data. We will present studies of invariant mass and transverse momentum resolution and the effect of acceptance and migration as a function of invariant mass. Both studies allow an estimate of key systematic uncertainties in the lepton energy scale that are used in the final analysis.