Event Filtering Research Articles

An Ensemble of Neural Networks for Online Electron Filtering at the ATLAS Experiment

The ATLAS experiment implemented an ensemble of neural networks (Neural-Ringer algorithm) dedicated to improving the performance of event filters selecting electrons in the high-input-rate online environment of the Large Hadron Collider (LHC) at CERN. This algorithm has been used online since 2017 to select electrons with transverse energies (ET) above 15 GeV. By taking advantage of calorimetry knowledge, the ensemble employs ring energy sums concentric to the electron candidate energy barycenter. The training procedure and final structure of the ensemble are designed to keep detector response flat with respect to particle energy and position. A detailed study was carried out to assess possible profile distortions in crucial offline quantities through the usage of statistical tests and residual analysis. NeuralRinger operation maintained high electron efficiency while improving fake rejection by a factor of 2 to 3, with negligible residuals in the offline quantities.

Track Reconstruction with PANDA at FAIR

The future PANDA experiment at FAIR aims to investigate a wide range of physics processes in antiproton-proton collisions at rates of up to 20 MHz, while employing a purely software-based event filter. To educate the trigger decisions, a full event reconstruction has to be carried out in real time. This challenge is amplified when considering tracks from particles with long lifetimes and displaced decay vertices, which add to the complexity of the reconstruction algorithms. Here, we present modifications to a cellular automatonbased track finder taking detector time-stamps into account in addition to spatial information, as well as several pattern recognition methods for longitudinal track reconstrucion with PANDA’s Straw Tube Tracker.

Perceptions of Classroom Teachers Regarding Values Education With Children's Games

Values are concepts that influence and direct individuals’ behaviors. Because values directly influence society’s structure, values education also takes an important place. Both children’s games and values education have a structure that can diversify values education and make it entertaining during learning. In this study, it was aimed to reveal the perceptions of classroom teachers regarding values education with children’s games. Within the scope of this aim, 272 classroom teachers were investigated for their perceptions. The study adopts a survey model and is of a descriptive quality. In the investigation of the study results, it was stated that teachers could conduct values education with children’s games and children’s games could be used in the educational environment as well as the fact that computer games make individuals antisocial. Additionally, the teachers who participated in the study stated that games ensured acting according to facts by filtering events and cases from a logical process, provided creative and innovative thinking and that teachers should know what games mean for children and how important is games’ influence in their development. The participating teachers also reported that children’s games include many elements of values that we brought from past to present and they reflected the cultural values of society.

Real-Time FPGA-Based Digital Signal Processing and Correction for a Small Animal PET

Small animal Positron Emission Tomography (PET) is dedicated to small animal imaging, which requires high position and energy precision, as well as good flexibility and efficiency of the electronics. This paper presents the design of a digital signal processing logic for a marmoset brain PET system based on LYSO crystal arrays, SiPMs, and the resistive network readout method. We implement 32-channel signal processing in a single Xilinx Artix-7 Field-Programmable Gate Array (FPGA). The logic is designed to support four online modes which are regular data processing mode, flood map construction mode, energy spectrum construction mode, and raw data mode. Several functions are integrated, including two-dimensional (2D) raw position calculation, crystal locating, events filtering, and synchronization detection. Furthermore, a series of online corrections is also integrated, such as photon peak correction to 511 keV and time measurement result correction with crystal granularity. A Gigabit Ethernet interface is utilized for data transfer, Look-Up Tables (LUTs) configuration, and command issuing. The pipeline logic works at 125 MHz with a signal processing capability beyond the required data rate of 1,000,000 events/s/channel. A series of initial tests are conducted. The results indicate that the logic design meets the application requirement.

Flit-Level InfiniBand Network Simulations of the DAQ System of the LHCb Experiment for Run-3

The LHCb (Large Hadron Collider beauty) experiment is designed to study differences between particles and anti-particles as well as very rare decays in the charm and beauty sector at the LHC (Large Hadron Collider). The detector will be upgraded in 2019 and a new trigger-less readout system will be implemented in order to significantly increase its efficiency and take advantage of the increased machine luminosity. In the upgraded system, both event building and event filtering will be performed in software for all the data produced in every bunch-crossing of the LHC. In order to transport the full data rate of 32 Tb/s we will use custom FPGA readout boards (PCIe40) and state of the art off-the-shelf network technologies. The full event building system will require around 500 nodes interconnected together. From a networking point of view, event building traffic has an all-to-all pattern, therefore it tends to create high network congestion. In order to maximize the link utilization different techniques can be adopted in various areas like traffic shaping, network topology and routing optimization. The size of the system makes it very difficult to test at production scale, before the actual procurement. We resort therefore to network simulations as a powerful tool for finding the optimal configuration. We will present an accurate low level description of an InfiniBand based network with event building like traffic. We will show comparison between simulated and real systems and how changes in the input parameters affect performances.

The LHCb DAQ Upgrade for LHC Run3

In order to optimize the physics reach in the large hadron collider (LHC) Run3, foreseen in the years 2021 and beyond, the LHC beauty experiment collaboration has decided to re-design its detector and the data acquisition system (DAQ). The new detector will operate at the LHC bunch-crossing frequency of 40 MHz without hardware trigger. The implementation of the DAQ and the online event filter farm (EFF) for the software trigger is challenging because of the expected data rate of about 40 Tb/s. At this scale, the DAQ and the EFF can be built at an affordable cost only by using off-the-shelf hardware as much as possible. In order to optimally exploit the fast evolution of technology, we need to choose the system architecture at the last moment. Therefore, it should not have strong dependences on specific technologies. We present here the system architecture, the different implementation options that we are studying along with measurements from these studies, and will explain the decision criteria and technology drivers for choosing the components for the final system.

Evaluating Kubernetes as an orchestrator of the Event Filter computing farm of the Trigger and Data Acquisition system of the ATLAS experiment at the Large Hadron Collider

The ATLAS experiment at the LHC relies on a complex and distributed Trigger and Data Acquisition (TDAQ) system to gather and select particle collision data. The Event Filter (EF) component of the TDAQ system is responsible for executing advanced selection algorithms, reducing the data rate to a level suitable for recording to permanent storage. The EF functionality is provided by a computing farm made up of thousands of commodity servers, each executing one or more processes. Moving the EF farm management towards a solution based on software containers is one of the main themes of the ATLAS TDAQ Phase-II upgrades in the area of the online software; it would make it possible to open new possibilities for fault tolerance, reliability and scalability. This paper presents the results of an evaluation of Kubernetes as a possible orchestrator of the ATLAS TDAQ EF computing farm. Kubernetes is a system for advanced management of containerized applications in large clusters. This paper will first highlight some of the technical solutions adopted to run the offline version of today’s EF software in a Docker container. Then it will focus on some scaling performance measurements executed with a cluster of 1000 CPU cores. In particular, this paper will report about the way Kubernetes scales in deploying containers as a function of the cluster size and show how a proper tuning of the Query per Second (QPS) Kubernetes parameter set can improve the scaling of applications in terms of running replicas. Finally, an assessment will be given about the possibility to use Kubernetes as an orchestrator of the EF computing farm in LHC’s Run 4.

Deep Learning Approach to Track Reconstruction in the upgraded VELO

The LHCb experiment will undergo a major upgrade for LHC Run III, scheduled to start taking data in 2021. In order to exploit the full physics potential of the upgraded detector, LHCb will employ a high level event filter entirely implemented in software. The event filter has to operate in real time at the 40 MHz LHC bunch crossing rate. The LHCb collaboration is currently exploring a variety of new new computing paradigms in order to cope with the challenges posed by the high data taking rates. One contribution to this effort is the application of Machine Learning (ML) techniques to particle track reconstruction. We explore an ML approach to the track reconstruction in the upgraded LHCb Vertex Locator (VELO). The reconstruction algorithm is evaluated with fully simulated Minimum-Bias data that reflects the input to the real time event filter. We have achieved a reasonable track reconstruction efficiency and low ghost rate with our first approach and are currently exploring several methods to further quality of the track reconstruction.

DyTER – Dynamic Track and Event Reconstruction

The future PANDA experiment at FAIR experiment aims to cover a wide range of processes in antiproton-proton collisions at event rates of up to 20 MHz. Such event rates make reconstruction a challenging task for the purely software-based event filter. Investigating complex event topologies with displaced vertices increases the difficulty even further. Here we present two attempts to meet these future challenges: an algorithm for track reconstruction based on pattern matching with pre-determined look-up tables, and as a continuation of this approach a system of neural networks for identifying specific particle track candidates and predicting their momentum.

Mining comorbidity patterns using retrospective analysis of big collection of outpatient records

BackgroundStudying comorbidities of disorders is important for detection and prevention. For discovering frequent patterns of diseases we can use retrospective analysis of population data, by filtering events with common properties and similar significance. Most frequent pattern mining methods do not consider contextual information about extracted patterns. Further data mining developments might enable more efficient applications in specific tasks like comorbidities identification.MethodsWe propose a cascade data mining approach for frequent pattern mining enriched with context information, including a new algorithm MIxCO for maximal frequent patterns mining. Text mining tools extract entities from free text and deliver additional context attributes beyond the structured information about the patients.ResultsThe proposed approach was tested using pseudonymised reimbursement requests (outpatient records) submitted to the Bulgarian National Health Insurance Fund in 2010–2016 for more than 5 million citizens yearly. Experiments were run on 3 data collections. Some known comorbidities of Schizophrenia, Hyperprolactinemia and Diabetes Mellitus Type 2 are confirmed; novel hypotheses about stable comorbidities are generated. The evaluation shows that MIxCO is efficient for big dense datasets.ConclusionExplicating maximal frequent itemsets enables to build hypotheses concerning the relationships between the exogeneous and endogeneous factors triggering the formation of these sets. MixCO will help to identify risk groups of patients with a predisposition to develop socially-significant disorders like diabetes. This will turn static archives like the Diabetes Register in Bulgaria to a powerful alerting and predictive framework.

Particle Identification on an FPGA Accelerated Compute Platform for the LHCb Upgrade

The current LHCb readout system will be upgraded in 2018 to a ‘triggerless’ readout of the entire detector at the LHC collision rate of 40 MHz. The corresponding bandwidth from the detector down to the foreseen dedicated computing farm (event filter farm), which acts as the trigger, has to be increased by a factor of almost 100 from currently 500 GBit/s up to 40 TBit/s. The event filter farm will pre-analyse the data and will select the events on an event by event basis. This will reduce the bandwidth down to a manageable size to write the interesting physics data to tape. The design of such a system is a challenging task, why different technologies are considered and have to be investigated for the different parts of the system. For the usage in the event building or in the event filter farm (trigger) an experimental FPGA accelerated computing platform is considered and therefore tested. FPGA compute accelerators are more and more used in standard servers like for Microsoft Bing search or Baidu search. The platform we use hosts a general CPU and an high performance FPGA linked via an high speed link. On the FPGA an accelerator is implemented. The used system is a two socket platform from Intel with a Xeon CPU and an FPGA. The CPU and the FPGA are connected via the point-to-point interconnect QPI, which is used to interconnect CPUs in industry standard server. The FPGA has cache-coherent memory access to the main memory of the server and can collaborate with the CPU. These cache-coherent architectures are better suited for real-time connections between FPGA and CPU as the usual PCIe FPGA accelerators. It is very likely that these platforms, which are built in general for high performance computing, are also very interesting for the High Energy Physics community. As First step it is tested to port the an existing LHCb RICH particle identification algorithm for the Cherenkov angle reconstruction to the experimental FPGA accelerated platform. We will compare the performance of this algorithm running on a Xeon CPU with the performance of the same algorithm, which is running on the Xeon-FPGA compute accelerator platform. This work is done in collaboration with Intel Corporation.

Using adaptive runtime filtering to support an event‐based performance analysis

SummaryEvent‐based performance monitoring and analysis are effective means when tuning parallel applications for optimal resource usage. In this article, we address the data capacity challenge that arises when applying the tracing methodology to large‐scale parallel applications and long execution times. Existing approaches use static, pre‐defined event filters to reduce the performance data to a manageable size. In contrast, we propose self‐guided filters that automatically adapt to an application's runtime behaviour and therefore, do not require any previous knowledge or application executions. Our contribution consists of four adaptive runtime filters, which target a specific type of data redundancy each. The filters focus on detecting identical events in loop iterations, constant events with no variation in time, and very short, highly frequent, typically not very meaningful events, having a severe impact on the total data volume. We evaluate our prototype implementation with five real‐world applications and achieve a data reduction of two orders of magnitude while increasing execution time less than 1%. Likewise, we show that the qualitative impact of our filters on performance analysis in state‐of‐the‐art analysis tools can be reduced by adding feedback methods and statistical information to the filtered traces. Copyright © 2017 John Wiley & Sons, Ltd.

Online detection and localisation of piglet crushing using vocalisation analysis and context data

Fatal piglet crushing by the mother sow is a pervasive economic and animal welfare issue in piglet production. To keep the mother sow in a farrowing cage is the established countermeasure. This facility is a compromise that results in an impairment of the sows’ welfare to the benefit of her piglets and the farmer. A natural behaviour pattern which is demonstrated by most but not all sows is to free the trapped piglet by a posture change. Promoting this behaviour through aversive stimulations is an alternative approach to reduce piglet mortality. This approach requires an identification and localisation of ongoing piglet trapping in real-time. The present study investigates the online analysis of piglet vocalisation for this purpose. The results show, that trapping related stress articulations are outnumbered by other stress related articulations by a factor of 1:140 in a farrowing compartment with only 4 sows. Theoretical calculations for larger compartments indicate that this ratio becomes even worse due to an increasing influence of vocalisation from neighbouring pens. However, the specificity could be increased to more than 95% and precision to approximately 30% while maintaining a sensitivity of approximately 70% by retrospectively applying context based event filters. This specificity would be sufficient to limit the average number of erroneous trapping detections to one detection per sow within 3days without a substantial loss of sensitivity. Effective parameters for filtering were the age of the piglets and the sows’ body posture history. Calculations with hypothetical spatial event filters showed that this classification performance could be maintained even in much larger farrowing compartments. Combined with an aversive stimulation principle that can be applied to a whole region, this detection technology could be useful to reduce piglet mortality in loose farrowing applications. An already known and effective stimulation principle of this type is floor vibration. Such an active piglet rescue system would allow limiting the impairment of welfare to only those sows that actually crush piglets and to the time when piglets are being crushed.

RFID Based Indoor Positioning System Using Event Filtering

Recently, location systems using RFID technology have been studied in indoor environments. However, the existing techniques require high computational cost to compute the location of a moving object because they compare the location proximity of all reference tags and objects. In this paper, we propose an RFID based location positioning scheme using event filtering, which reduces the computation cost of calculating the locations of moving objects while maintaining the accuracy of location estimation. In addition, we propose an incremental location update policy to reduce the location update cost for moving objects. We also compare the proposed scheme with one of the localization schemes, LANDMARC using a performance evaluation. As a result, the proposed scheme outperforms LANDMARC in terms of the computational cost of location estimation. The proposed scheme also reduces the cost of location update by using the RFID-based update policy.

Technical note: The use of an accelerometer for measuring step activity and lying behaviors in dairy calves.

Calf behaviors such as step activity, lying bouts, and lying time may be an indicator of calf health and welfare. To reduce time-consuming visual observations, the use of behavioral monitoring systems have been developed to capture these data. Previous studies have validated lying behaviors using an accelerometer (HPG; HOBO Pendant G data logger, Onset Computer Corp., Bourne, MA) in calves. However, the HPG does not measure step activity. The objectives of this study were to (1) validate step activity, lying bouts, and lying time of AfiTag II (AT2; AfiTag II, Afimilk Ltd., Kibbutz Afikim, Israel) to observations from video, and (2) to compare the behavioral data from AT2 to the HPG. Calves (n=5) were group housed with an automatic calf feeder. Video cameras were installed at both sides of the pen, and observations were analyzed for 7h/calf. The AT2 and the HPG were both attached to the lateral side of the right rear leg of 5 calves, and data were recorded for 10 d. The full 10-d data set was used to examine correlations for lying bouts and lying time between AT2 and the HPG. The HPG was set at a 60-s sampling interval and the output was analyzed both unfiltered as well as utilizing a 1-min event filter to remove potentially erroneous readings. The AT2 recorded step activity, lying bouts, and lying time, and summarized these behaviors in 15-min periods. The AT2 recorded lying time in 3-min intervals, which were then automatically summarized in 15-min periods. The correlations of step activity, lying bouts, and lying time between video recordings and AT2 were 0.99. For the second objective, correlations between AT2 and the HPG were 0.99 for lying time and 0.93 for lying bouts. The 1-min event filter resulted in a 0.03 improvement in correlations for lying bouts between the HPG and AT2. The high correlation between video recordings and AT2 suggest that this device can be used to measure step activity, lying time, and lying bouts in unweaned dairy calves housed in groups.

Advanced boundary virtual reference algorithm for an indoor system using an active RFID interrogator and transponder

Localization of persons and objects has become a necessity in many industrial services. In indoor circumstances, radio frequency identification (RFID) has superior performance. However, radio signals within indoor environments are generally weak, and the tags have very restricted abilities. In addition, the multipath propagation, total cost, and signal interference increase with increasing number of reference tags over a certain limit. Therefore, to improve the performance of indoor positioning for real-time localization, a new active acquisition method for an active RFID system that works at 2.4 GHz has been proposed; it is called the boundary virtual reference (BVIRE) algorithm. It depends on boundary virtual reference tags rather than increasing the number of real reference tags, which in turn reduces the total cost while maintaining the location accuracy. We have implemented a linear regression method to further enhance the positioning accuracy while eliminating the unnecessary tags from the estimation method using event filtering, in which only a few neighbouring reference tags are helpful in deciding the location of the object tag. The experimental results show that our BVIRE algorithm considerably reduces the error estimation compared with previous algorithms. The system has enhanced the positioning accuracy and lowered the total costs. In addition, the localization precision of the proposed approach has been significantly increased to approximately 90.25 % compared with PinPoint algorithm with no additional reference tags or radio frequency interference; this represents a significant improvement over other algorithms.

The ATLAS Trigger System: Past, Present and Future

The ATLAS trigger has been used very successfully for the online event selection during the first run of the LHC between 2009-2013 at a centre-of-mass energy between 900 GeV and 8 TeV. The trigger system consists of a hardware Level-1 (L1) and a software based high-level trigger (HLT) that reduces the event rate from the design bunch-crossing rate of 40 MHz to an average recording rate of a few hundred Hz.We will briefly review the performance of the ATLAS trigger system during the past data-taking period and point out the challenges for the trigger system during the next LHC run in early 2015 with a smaller bunch spacing, almost twice the centre-of-mass energy and higher peak luminosity. We will show the ongoing improvements and upgrades to the existing system that will ensure an even better performing trigger system despite the harsher machine conditions. This includes changes to the L1 calorimeter trigger, the introduction of a new L1 topological trigger module, improvements in the L1 muon system and the merging of the previously two-level HLT system into a single event filter farm. In addition, we will give an overview of the algorithmic improvements in the various HLT algorithms used to identify leptons, hadrons and global event quantities like missing transverse energy.

LHCb GPU acceleration project

The LHCb detector is due to be upgraded for processing high-luminosity collisions, which will increase data bandwidth to the event filter farm from 100 GB/s to 4 TB/s, encouraging us to look for new ways of accelerating Online reconstruction. The Coprocessor Manager is a new framework for integrating LHCb's existing computation pipelines with massively parallel algorithms running on GPUs and other accelerators. This paper describes the system and analyzes its performance.

Performance benchmark of LHCb code on state-of-the-art x86 architectures

For Run 2 of the LHC, LHCb is replacing a significant part of its event filter farm with new compute nodes. For the evaluation of the best performing solution, we have developed a method to convert our high level trigger application into a stand-alone, bootable benchmark image. With additional instrumentation we turned it into a self-optimising benchmark which explores techniques such as late forking, NUMA balancing and optimal number of threads, i.e. it automatically optimises box-level performance. We have run this procedure on a wide range of Haswell-E CPUs and numerous other architectures from both Intel and AMD, including also the latest Intel micro-blade servers. We present results in terms of performance, power consumption, overheads and relative cost.

60 GHz wireless data transfer for tracker readout systems—first studies and results

To allow highly granular trackers to contribute to first level trigger decisions or event filtering, a fast readout system with very high bandwidth is required. Space, power and material constraints, however, pose severe limitations on the maximum available bandwidth of electrical or optical data transfers. A new approach for the implementation of a fast readout system is the application of a wireless data transfer at a carrier frequency of 60 GHz. The available bandwidth of several GHz allows for data rates of multiple Gbps per link. 60 GHz transceiver chips can be produced with a small form factor and a high integration level. A prototype transceiver currently under development at the University of Heidelberg is briefly described in this paper. To allow easy and fast future testing of the chip's functionality, a bit error rate test has been developed with a commercially available transceiver. Crosstalk might be a big issue for a wireless readout system with many links in a tracking detector. Direct crosstalk can be avoided by using directive antennas, linearly polarized waves and frequency channeling. Reflections from tracking modules can be reduced by applying an absorbing material like graphite foam. Properties of different materials typically used in tracking detectors and graphite foam in the 60 GHz frequency range are presented. For data transmission tests, links using commercially available 60 GHz transmitters and receivers are used. Studies regarding crosstalk and the applicability of graphite foam, Kapton horn antennas and polarized waves are shown.