Corrigendum: Applications and techniques for fast machine learning in science.

[This corrects the article DOI: 10.3389/fdata.2022.787421.].

Applications and Techniques for Fast Machine Learning in Science.

In this community review report, we discuss applications and techniques for fast machine learning (ML) in science-the concept of integrating powerful ML methods into the real-time experimental data processing loop to accelerate scientific discovery. The material for the report builds on two workshops held by the Fast ML for Science community and covers three main areas: applications for fast ML across a number of scientific domains; techniques for training and implementing performant and resource-efficient ML algorithms; and computing architectures, platforms, and technologies for deploying these algorithms. We also present overlapping challenges across the multiple scientific domains where common solutions can be found. This community report is intended to give plenty of examples and inspiration for scientific discovery through integrated and accelerated ML solutions. This is followed by a high-level overview and organization of technical advances, including an abundance of pointers to source material, which can enable these breakthroughs.

Graph Neural Networks for Charged Particle Tracking on FPGAs.

The determination of charged particle trajectories in collisions at the CERN Large Hadron Collider (LHC) is an important but challenging problem, especially in the high interaction density conditions expected during the future high-luminosity phase of the LHC (HL-LHC). Graph neural networks (GNNs) are a type of geometric deep learning algorithm that has successfully been applied to this task by embedding tracker data as a graph-nodes represent hits, while edges represent possible track segments-and classifying the edges as true or fake track segments. However, their study in hardware- or software-based trigger applications has been limited due to their large computational cost. In this paper, we introduce an automated translation workflow, integrated into a broader tool called hls4ml, for converting GNNs into firmware for field-programmable gate arrays (FPGAs). We use this translation tool to implement GNNs for charged particle tracking, trained using the TrackML challenge dataset, on FPGAs with designs targeting different graph sizes, task complexites, and latency/throughput requirements. This work could enable the inclusion of charged particle tracking GNNs at the trigger level for HL-LHC experiments.

Design of an FPGA-Based Algorithm for Real-Time Solutions of Statistics-Based Positioning.

We report on the implementation of an algorithm and hardware platform to allow real-time processing of the statistics-based positioning (SBP) method for continuous miniature crystal element (cMiCE) detectors. The SBP method allows an intrinsic spatial resolution of ~1.6 mm FWHM to be achieved using our cMiCE design. Previous SBP solutions have required a postprocessing procedure due to the computation and memory intensive nature of SBP. This new implementation takes advantage of a combination of algebraic simplifications, conversion to fixed-point math, and a hierarchal search technique to greatly accelerate the algorithm. For the presented seven stage, 127 × 127 bin LUT implementation, these algorithm improvements result in a reduction from >7 × 10(6) floating-point operations per event for an exhaustive search to < 5 × 10(3) integer operations per event. Simulations show nearly identical FWHM positioning resolution for this accelerated SBP solution, and positioning differences of <0.1 mm from the exhaustive search solution. A pipelined field programmable gate array (FPGA) implementation of this optimized algorithm is able to process events in excess of 250 K events per second, which is greater than the maximum expected coincidence rate for an individual detector. In contrast with all detectors being processed at a centralized host, as in the current system, a separate FPGA is available at each detector, thus dividing the computational load. These methods allow SBP results to be calculated in real-time and to be presented to the image generation components in real-time. A hardware implementation has been developed using a commercially available prototype board.

FPGA-Based Front-End Electronics for Positron Emission Tomography.

Modern Field Programmable Gate Arrays (FPGAs) are capable of performing complex discrete signal processing algorithms with clock rates above 100MHz. This combined with FPGA's low expense, ease of use, and selected dedicated hardware make them an ideal technology for a data acquisition system for positron emission tomography (PET) scanners. Our laboratory is producing a high-resolution, small-animal PET scanner that utilizes FPGAs as the core of the front-end electronics. For this next generation scanner, functions that are typically performed in dedicated circuits, or offline, are being migrated to the FPGA. This will not only simplify the electronics, but the features of modern FPGAs can be utilizes to add significant signal processing power to produce higher resolution images. In this paper two such processes, sub-clock rate pulse timing and event localization, will be discussed in detail. We show that timing performed in the FPGA can achieve a resolution that is suitable for small-animal scanners, and will outperform the analog version given a low enough sampling period for the ADC. We will also show that the position of events in the scanner can be determined in real time using a statistical positioning based algorithm.

FPGA-Based Pulse Parameter Discovery for Positron Emission Tomography.

Modern Field Programmable Gate Arrays (FPGAs) are capable of performing complex digital signal processing algorithms with clock rates well above 100MHz. This, combined with FPGA's low expense and ease of use make them an ideal technology for a data acquisition system for a positron emission tomography (PET) scanner. The University of Washington is producing a series of high-resolution, small-animal PET scanners that utilize FPGAs as the core of the front-end electronics. For these next generation scanners, functions that are typically performed in dedicated circuits, or offline, are being migrated to the FPGA. This will not only simplify the electronics, but the features of modern FPGAs can be utilizes to add significant signal processing power to produce higher resolution images. In this paper we report how we utilize the reconfigurable property of an FPGA to self-calibrate itself to determine pulse parameters necessary for some of the pulse processing steps. Specifically, we show how the FPGA can generate a reference pulse based on actual pulse data instead of a model. We also report how other properties of the photodetector pulse (baseline, pulse length, average pulse energy and event triggers) can be determined automatically by the FPGA.

Design of a Real Time FPGA-based Three Dimensional Positioning Algorithm.

We report on the implementation and hardware platform of a real time Statistics-Based Processing (SBP) method with depth of interaction processing for continuous miniature crystal element (cMiCE) detectors using a sensor on the entrance surface design. Our group previously reported on a Field Programmable Gate Array (FPGA) SBP implementation that provided a two dimensional (2D) solution of the detector's intrinsic spatial resolution. This new implementation extends that work to take advantage of three dimensional (3D) look up tables to provide a 3D positioning solution that improves intrinsic spatial resolution. Resolution is most improved along the edges of the crystal, an area where the 2D algorithm's performance suffers. The algorithm allows an intrinsic spatial resolution of ~0.90 mm FWHM in X and Y and a resolution of ~1.90 mm FWHM in Z (i.e., the depth of the crystal) based upon DETECT2000 simulation results that include the effects of Compton scatter in the crystal. A pipelined FPGA implementation is able to process events in excess of 220k events per second, which is greater than the maximum expected coincidence rate for an individual detector. In contrast to all detectors being processed at a centralized host, as in the current system, a separate FPGA is available at each detector, thus dividing the computational load. A prototype design has been implemented and tested using a reduced word size due to memory limitations of our commercial prototyping board.

Simulation of Algorithms for Pulse Timing in FPGAs.

Modern Field Programmable Gate Arrays (FPGAs) are capable of performing complex discrete signal processing algorithms with clock rates well above 100MHz. This, combined with FPGA's low expense and ease of use, make them an ideal technology for pulse timing and are a central part of our next generation of electronics for our pre-clinical PET scanner systems. To that end, our laboratory has been developing a pulse timing technique that uses pulse fitting to achieve timing resolution well below the sampling period of the analog to digital converter (ADC). While ADCs with sampling rates in excess of 400MS/s exist, we feel that using ADCs with lowing sampling rates has many advantages for positron emission tomography (PET) scanners. It is with this premise that we have started simulating timing algorithms using MATLAB in order to optimize the parameters before implementing the algorithm in Verilog. MATLAB simulations allow us to quickly investigate filter designs, ADC sampling rates and algorithms with real data before implementation in hardware. We report our results for a least squares fitting algorithm and a new version of a leading edge detector of PMT pulses.