Michael Waßmer

In high-energy particle collisions, the primary collision products usually decay further resulting in tree-like, hierarchical structures with a priori unknown multiplicity. At the stable-particle level all decay products of a collision form permutation invariant sets of final state objects. The analogy to mathematical graphs gives rise to the idea that graph neural networks (GNNs), which naturally resemble these properties, should be best-suited to address many tasks related to high-energy particle physics. In this paper we describe a benchmark test of a typical GNN against neural networks of the well-established deep fully-connected feed-forward architecture. We aim at performing this comparison maximally unbiased in terms of nodes, hidden layers, or trainable parameters of the neural networks under study. As physics case we use the classification of the final state X produced in association with top quark-antiquark pairs in proton-proton collisions at the Large Hadron Collider at CERN, where X stands for a bottom quark-antiquark pair produced either non-resonantly or through the decay of an intermediately produced Z or Higgs boson.

Beginning with the first event in Washington, DC in 2002, the Department of Energy's Solar Decathlon has brought attention to the promise of PV-powered, zero-energy homes through the format of a riveting collegiate competition. As an internationally recognized event, it demonstrates innovative solutions-using energy efficiently and generating the needs of modern domestic life with solar energy. At the 2007 event (the third Solar Decathlon), a new era of the competition was conceived, in which the U.S. Department of Energy's event (produced and managed by the National Renewable Energy Laboratory) would be complemented by a European Solar Decathlon in Madrid, with each competition to be held in alternating years. Each of these upcoming events is described.

Solar Decathlon 2005: The Event in Review is a technical report describing the 2005 Solar Decathlon, an event sponsored by the U.S. Department of Energy wherein 18 collegiate teams competed in 10 contests to design, build, and operate an attractive, efficient, entirely solar-powered home. The report gives an overview of the competition, including final results, team strategies, and detailed descriptions the 18 homes.

The Solar Decathlon is a college-level student competition to design, construct, and operate approximately 800-ft/sup 2/ homes that are highly energy efficient and use solar thermal and photovoltaic technologies to meet all household energy needs. In fall 2005, 19 teams compete for about one week in ten contests, six of which test energy usage and production. The competition takes place on the National Mall in Washington, D.C., and is open to the public. A comprehensive building-energy simulation and monitoring project developed for Solar Decathlon homes is outlined in this paper. Simulation, monitoring, and analysis activities are conducted before, during, and after the competition in five phases. The level of detail of monitoring activities progressively increases in each phase. The level of detail of the simulations remains consistent throughout, but the simulation accuracy is expected to improve as more detailed monitoring data become available for model calibration. The results of the various phases of the project are expected to be of interest to five diverse groups: visitors to the Solar Decathlon, competition organizers, current Solar Decathlon teams, future Solar Decathlon teams, and building scientists.

The 2005 Solar Decathlon, an 18-team international collegiate competition of solar-powered homes, was held on the National Mall, Washington, D.C., in October 2005. For the teams, accomplishing the energy-consuming tasks required by the 10-contest competition and public event was quite challenging during the particular week of the competition owing to unseasonably cloudy and rainy weather. While no team had designed for as little solar radiation as was available during those five competition days, successful teams employed a combination of technical acumen and savvy strategy to accumulate the most points in the contests scored by objective measurements. (Some contests were scored subjectively by judges.) Following are a description of the competition, a description of the solar power systems used by entries in the Solar Decathlon, a summary of how the teams operated the PV systems in their homes, and an assessment of how the homes actually performed in the measured contests during the 2005 competition. The potential for operation of non-Solar Decathlon households with PV systems employing battery backup during periods of grid outage and limited solar resource is also discussed

In 2002, the Department of Energy (DOE) sponsored the world’s first university competition to design and build a completely solar powered house. One requirement of the competition was to perform simulations of the house’s photovoltaic, solar thermal, and space conditioning systems. By instituting this requirement, DOE is encouraging the building industry to apply the “whole-building design” approach to residences as a method of reducing financial and environmental operation costs of the building over its lifetime. This paper describes the simulation approach taken by the University of Colorado Solar Decathlon Team. In addition to describing the process of simulating a zero-energy residential building, the specific results of the simulations and related parametric studies are also presented. The design and analysis process provides a case study in the application of six different simulation tools for zero-energy building design. Energy-10 provided an environment for parametric analysis of building design options during the critical early design phase. However, it lacks the flexibility to model solar electric, solar thermal, and specialized HVAC systems. FChart gave valuable guidance early in the project on the impact of solar system sizing and performance. TRNSYS is extremely flexible in that it can simulate various solar systems and the interactions of virtually any thermal system commonly found in buildings. This flexibility is accompanied by the burden of complexity and a generic user interface that limits its use as a routine building design tool. Radiance, AGI32, and ECOTECT provided specialized simulation tools for the integration of the daylight delivery system, external shading devices, and the electric lighting system. Additional development is required to better integrate these design needs into general building energy analysis tools.

In the 2005 Solar Decathlon competition, 19 collegiate teams will design, build, and operate grid-independent homes powered by photovoltaic (PV) arrays on the National Mall. The prominence of grid-interconnected systems in the marketplace has provided the impetus for the development of a net-metering exhibit to be installed and operated during the competition. The exhibit will inform the visiting public about PV basics and appropriate alternatives to grid-independent systems. It will consist of four interactive components. One will be designed to educate people about the principles of net metering using a small PV array, a grid-interactive inverter, and a variable load. Additional components of the exhibit will demonstrate the effects of orientation, cloud cover, and nighttime on performance. The nighttime component will discuss appropriate storage options for different applications.

A search for the production of Dark Matter in hadronic mono-top signatures using data recorded by the CMS experiment in the years from 2016 to 2018 is presented. First, the analysis strategy is defined relying on the existence of large missing transverse momentum and a large-radius jet. The underlying theoretical model is then studied in order to validate the analysis strategy. Consequently, the modeling of the production of single vector bosons in association with additional jets is studied in detail because of its importance regarding the background model of this analysis. Corrections based on available state-of-the-art theoretical calculations are derived in order to improve the modeling of the aforementioned processes and these corrections are compared to alternative corrections based on Monte Carlo simulation. Next, the data analysis and its technical details are explained. Then, a statistical model is built estimating the major background processes utilizing data from control regions. Finally, a statistical hypothesis test combined with parameter estimation is performed in order to search for possible signal contributions in data. No significant signal contributions were found. Therefore, parameter ranges of the used theoretical signal model were excluded.