American Fuzzy Lop (AFL) is an evolutionary fuzzer that is strategically implemented as a tool for discovering bugs in software during vulnerability research. This work seeks to understand how to best implement AFL on the High-Performance Computing resources available on the unclassified network at Sandia National Laboratories. We investigate various methods of executing AFL, requesting varying numbers of tasks on single compute nodes with 36 physical cores and 72 total threads. A Python script called Blue Claw is presented as an automated testbed generator tool to assist in the tedious process of creating and executing experiments of any scale and duration.
Geotechnical concerns arise due to the close proximity of the some of the caverns to each other (e.g., Caverns 15 and 17) or to the edge of the salt dome (e.g., Cavern 20). There are nine abandoned caverns, one of which collapsed (Cavern 7) in 1954 and another (Cavern 4) which is believed to be in a quasi-stable condition. This report provides explanations for these geotechnical concerns. The structural integrity of the pillar between BC-15 and 17 is examined. No salt fall is expected through 2045. However, the dilatant damaged area increases with time, especially, at the chimney area of BC-17. One drawdown leach for both caverns could be allowed if they are normally operated as a gallery, depressurized simultaneously. The possibility of a loss in integrity of BC-20 is examined in the salt between the dome edge and the cavern. The edge pillar is predicted to have experienced tensile stress since September 1999, but the small tensile stressed area is predicted to disappear in 2018 because BC-20 is filled fully with brine rather than oil since 2/7/2013. Even though BC-20 is no longer used as an SPR cavern, we need to continue monitoring the cavern integrity. BC-4 is also currently filled with brine and will not hold pressure at the wellhead. The cavern extends upward into the caprock and has no effective salt roof The results indicate that any sort of caprock roof collapse for BC-4 is not imminent but salt falls will likely occur from the near-roof portions of the cavern. The uncertainty due to salt falls illustrates the importance of continued monitoring of the area around BC-4 for behavior such as subsidence and tilt which may indicate a change in the cavern's integrity status.
In this study we investigate how an ensemble of disease models can be conditioned to observational data, in a bid to improve its predictive skill. We use the ensemble of influenza forecasting models gathered by the US Centers for Disease Control and Prevention (CDC) as the exemplar. This ensemble is used every year to forecast the annual influenza outbreak in the United States. The models constituting this ensemble draw on very different modeling assumptions and approximations and are a diverse collection of methods to approximate epidemiological dynamics. Currently, each models' predictions are accorded the same importance, or weight, when compiling the ensemble's forecast. We consider this equally-weighted ensemble as the baseline case which has to be improved upon. In this study, we explore whether an ensemble forecast can be improved by "conditioning" the ensemble to whatever observational data is available from the ongoing outbreak. "Conditioning" can imply according the ensemble's members different weights which evolve over time, or simply perform the forecast using the top k (equally-weighted) models. In the latter case, the composition of the "top-k-see of models evolves over time. This is called "model averaging" in statistics. We explore four methods to perform model-averaging, three of which are new. We find that the CDC ensemble responds best to the "top-k-models" approach to model-averaging. All the new MA methods perform better than the baseline equally-weighted ensemble. The four model-averaging methods treat the models as black-boxes and simply use their forecasts as inputs i.e., one does not need access to the models at all, but rather only their forecasts. The model-averaging approaches reviewed in this report thus form a general framework for model-averaging any model ensemble.