-
Commissioning of the laser-driven ion acceleration beamline at the Centre for Advanced Laser Applications
Authors:
Jens Hartmann,
Thomas F. Rösch,
Felix Balling,
Marc Berndl,
Leonard Doyle,
Lotta Flaig,
Sonja Gerlach,
Luisa Tischendorf,
Jörg Schreiber
Abstract:
The Centre for Advanced Laser Applications (CALA) in Garching near Munich features the ATLAS 3000 laser system, which can deliver up to 3\,PW within a pulse length of 20\,fs. It is the driver for the Laser-driven ION (LION) beamline, which aims to accelerate protons and carbons for applications. For commissioning, we currently operate with 5\,J on target in 28\,fs. A $20\degree$ off-axis parabolic…
▽ More
The Centre for Advanced Laser Applications (CALA) in Garching near Munich features the ATLAS 3000 laser system, which can deliver up to 3\,PW within a pulse length of 20\,fs. It is the driver for the Laser-driven ION (LION) beamline, which aims to accelerate protons and carbons for applications. For commissioning, we currently operate with 5\,J on target in 28\,fs. A $20\degree$ off-axis parabolic mirror focuses the 28\,cm diameter laser-beam down to a micrometer-sized spot, where a vacuum-compatible wave-front sensor is used in combination with a deformable mirror for focus optimization. The nano-Foil Target Positioning System (nFTPS) can replace targets with a repetition rate of up to 0.5\,Hz and store up to 19 different target foils. A dipole magnet in a wide-angle spectrometer configuration deflects ions onto a CMOS detector for an online read-out. Commissioning started mid 2019 with regular proton acceleration using nm-thin plastic foils as targets. Since then proton cut-off energies above 20\,MeV have been regularly achieved. The amount of light traveling backwards from the experiment into the laser is constantly monitored and 5\,J on target have been determined as the current limit to prevent damage in the laser. Protons with a kinetic energy of 12\,MeV are stably accelerated with the given laser parameters and are suitable for transport with permanent magnet quadrupoles towards our application platform. We have performed parameter scans varying target thicknesses to optimize for highest and most stable proton numbers at 12\,MeV kinetic energy, and investigated shot-to-shot particle number stability for the best parameters.
△ Less
Submitted 15 November, 2021;
originally announced November 2021.
-
It's easy to fool yourself: Case studies on identifying bias and confounding in bio-medical datasets
Authors:
Subhashini Venugopalan,
Arunachalam Narayanaswamy,
Samuel Yang,
Anton Geraschenko,
Scott Lipnick,
Nina Makhortova,
James Hawrot,
Christine Marques,
Joao Pereira,
Michael Brenner,
Lee Rubin,
Brian Wainger,
Marc Berndl
Abstract:
Confounding variables are a well known source of nuisance in biomedical studies. They present an even greater challenge when we combine them with black-box machine learning techniques that operate on raw data. This work presents two case studies. In one, we discovered biases arising from systematic errors in the data generation process. In the other, we found a spurious source of signal unrelated…
▽ More
Confounding variables are a well known source of nuisance in biomedical studies. They present an even greater challenge when we combine them with black-box machine learning techniques that operate on raw data. This work presents two case studies. In one, we discovered biases arising from systematic errors in the data generation process. In the other, we found a spurious source of signal unrelated to the prediction task at hand. In both cases, our prediction models performed well but under careful examination hidden confounders and biases were revealed. These are cautionary tales on the limits of using machine learning techniques on raw data from scientific experiments.
△ Less
Submitted 6 April, 2020; v1 submitted 12 December, 2019;
originally announced December 2019.
-
Molecular Graph Convolutions: Moving Beyond Fingerprints
Authors:
Steven Kearnes,
Kevin McCloskey,
Marc Berndl,
Vijay Pande,
Patrick Riley
Abstract:
Molecular "fingerprints" encoding structural information are the workhorse of cheminformatics and machine learning in drug discovery applications. However, fingerprint representations necessarily emphasize particular aspects of the molecular structure while ignoring others, rather than allowing the model to make data-driven decisions. We describe molecular "graph convolutions", a machine learning…
▽ More
Molecular "fingerprints" encoding structural information are the workhorse of cheminformatics and machine learning in drug discovery applications. However, fingerprint representations necessarily emphasize particular aspects of the molecular structure while ignoring others, rather than allowing the model to make data-driven decisions. We describe molecular "graph convolutions", a machine learning architecture for learning from undirected graphs, specifically small molecules. Graph convolutions use a simple encoding of the molecular graph---atoms, bonds, distances, etc.---which allows the model to take greater advantage of information in the graph structure. Although graph convolutions do not outperform all fingerprint-based methods, they (along with other graph-based methods) represent a new paradigm in ligand-based virtual screening with exciting opportunities for future improvement.
△ Less
Submitted 18 August, 2016; v1 submitted 2 March, 2016;
originally announced March 2016.
