12 October 2020 – The KM3NeT Collaboration has published a new paper that aims at demonstrating the general applicability of deep convolutional neural networks to neutrino telescopes, using simulated datasets for the KM3NeT/ORCA detector as an example. To this end, the networks are employed to achieve reconstruction and classification tasks that constitute an alternative to the analysis pipeline presented for KM3NeT/ORCA in the KM3NeT Letter of Intent. They are used to infer event reconstruction estimates for the energy, the direction, and the interaction point of incident neutrinos. The spatial distribution of Cherenkov light generated by charged particles induced in neutrino interactions is classified as shower- or track-like, and the main background processes associated with the detection of atmospheric neutrinos are recognized. Performance comparisons to machine-learning classification and maximum-likelihood reconstruction algorithms previously developed for KM3NeT/ORCA are provided.
The conclusion is that this application of deep convolutional neural networks to simulated datasets for a large-volume neutrino telescope yields competitive reconstruction results and performance improvements with respect to classical approaches.
Event reconstruction for KM3NeT/ORCA using convolutional neural networks
10 September 2020 – Like in all oceans, deep in the Mediterranean Sea turbulent waves occur which influence under water life because they transport water of different temperatures and important nutrition. Understanding the occurrence and behaviour of the ‘underwater waves’ is the objective of Hans van Haren and his team of the Royal Netherlands Institute for Sea Research NIOZ – member of the KM3NeT Collaboration (see article in Europhysics News 51/2 ).
The team has developed a large mooring construction comprising an array of temperature sensors that during two years will precisely measure the temperature of the deep sea water near the KM3NeT site off-shore Toulon, France. The measured temperature profiles will reveal the existence and behaviour of underwater turbulence and internal waves at the site. From a distance the image of the mooring resembles that of the KM3NeT array of optical sensors (see picture below).
The mooring consists of a 70 m diameter large steel ring, holding a network with 3000 high-precision temperature sensors distributed over 45 vertical lines, 125 m high and 9.5 m apart. On land this already looks quite impressive (see drone video below) but in sea the whole construction will fill a half cubic hectometre seawater volume. The installation of the mooring has some resemblance with the installation of the sensor array of KM3NeT. This is not surprising since the NIOZ is the institute where the KM3NeT compact deployment method was invented first. The lines with temperature sensors are compacted in small packets that are anchored on the seabed. Then they unfurl one by one to their full lengths. The major difference with the KM3NeT deployment technique is that the mooring structure of 45 lines is deployed as a whole, while for KM3NeT each line is deployed separately.
The deployment of the temperature mooring is planned for the second week of October after assembly in the harbor of Toulon. Stay tuned!
See also : a drone video by Hung-An Tian, NIOZ PhD student at
Pictures (courtesy NIOZ): After assembly, the mooring is towed to the deployment site and deployed using a custom-made ‘parachute’. Once in position, the lines will automatically unroll to to their full length after five days.
UPDATE on 2020/10/02
The NIOZ 3D temperature array is being assembled in the harbour of La Seyne sur Mer between 28 September and 6 October (see pictures). Today the construction had to be abandoned due to a rain-storm.
After Covid-19 tests and quarantine of the team, the RV Pelagia of the NIOZ Institute will sail out and tow the structure to its location for deployment sometime between 9 and 15 October, depending on permissions and the weather conditions.
Despite the devastating storm and heavy rain of the last few days in France, the assembly of the array is almost ready.
UPDATE on 2020/10/09
The RV Pelagia of the Royal NIOZ institute in the Netherlands has arrived in La Seyne dur Mer and departed again towing the large structure for the deployment site near the KM3NeT site.
At the moment of writing this update, the Pelagia has arrived at the deployment site, about 40 kilometer off-shore and is maneuvering (see screen shot of the Marine Traffic site).
UPDATE on 2020/10/10
Preparations for the deployment of the structure, which in the mean time has reached its designated position at the seabed at a depth of about 2.5 km. It will stay there for 3 years.
Update on 23/11/2020
A ROV visited the structure at a depth of about 2400 m. The pictures made by the ROV show buoys at the top of the moorings and part of the steel ring on the seabed together with a few anchors of the moorings. The temperature sensors are visible as small tubes attached to the cables.
On 27 July 2020, the ORCA detector of KM3NeT reached a milestone: its first 6 strings were continuously taking data since 6 months. With two musical productions of the amazing talents in the KM3NeT Collaboration, the milestone was celebrated.
Enjoy ‘6 strings, 6 months’, the song of the Route 66 of KM3NeT and an instrumental piece on 6 pianos by 6 players.
Both productions were recorded in corona times – at large distances between the performers.
13 July 2020 – The KM3NeT Collaboration has published the details of gSeaGen, a simulation software package for efficient generation of neutrino events for the analysis of measured light signals in the KM3NeT telescopes. Monte Carlo simulations play an important role in the data analysis of neutrino telescopes. They are used to design reconstruction algorithms for neutrino events and to estimate cosmic and atmospheric signals in various physics analyses.
The new gSeaGen software tool is based on code of the GENIE Collaboration which aims at developing a global software platform for the Monte Carlo simulation of neutrino interactions with energies up to PeV scales. Currently, the GENIE simulation code focuses mainly on events in the low-energy range (≤5 GeV) and is valid up to 5 TeV.
As described in the paper, the gSeaGen tool allows for the generation of electron, muon and tau neutrino. Its application for the KM3NeT telescopes is described in detail.
KM3NeT Collaboration, S. Aiello, et al., Computer Physics Communications 256 (2020) 107477
17 June 2020 – The KM3NeT Collaboration has published a new paper about the control unit of the data acquisition system. The data acquisition control software of KM3NeT is operating both the off-shore detectors in the deep sea and in the lab the testing and qualification stations for detector components. The software, named Control Unit, is highly modular. It can undergo upgrades and reconfiguration with the acquisition running. Interplay with the central database of the Collaboration is obtained in a way that allows for data taking even if Internet links fail. In order to simplify the management of computing resources in the long term, and to cope with possible hardware failures of one or more computers, the KM3NeT Control Unit software features a custom dynamic resource provisioning and failover technology, which is especially important for ensuring continuity in case of rare transient events in multi-messenger astronomy. The software architecture relies on ubiquitous tools and broadly adopted technologies and has been successfully tested on several operating systems.
10 June 2020 – The KM3NeT Collaboration is deeply saddened by the recent outbreaks of violence and hatred against people of colour. They once again laid bare the enduring worldwide systemic racism.
The researchers in KM3NeT are strongly against any kind of racism or discrimination. We urge all citizens of the world and their leaders to embrace all actions suited to establish equal opportunities for all, and forever.
As a collaboration, we will increase awareness on the impact of unintended racism and discrimination in our universities and research institutes and in particular in our collaboration.
8 June 2020 – Like so many other meetings, also the Spring Collaboration meeting of KM3NeT went online during corona times. A week full of discussions started today. An online concert and quiz are planned. Of course the traditional group photo has already been made.
05 February 2020 – The ORCA6 detector of KM3NeT is taking data since 27 January 2020 on a 24/7 scheme. Physicists are ‘on-shift’ to remotely or on-site operate the detector in the deep sea. The recorded data is stored in the computer centres of the KM3NeT Collaboration for further analysis.
The first step is to reconstruct from the recorded light flashes the path of charged particles through the ORCA6 detector. Most of them are muon particles generated in the Earth’s atmosphere and travelling through the detector from above. We showed already an example in the news item of 27 January.
In the video below we show a series of five charged particles entering the detector from below or from the side. This is an indication that they have been created in an interaction of a neutrino with the matter surrounding the detector.
In the picture below, you see the plots that KM3NeT physicists like: six plots showing for each of the six detection units in ORCA6 the optical sensors that – in the pitch dark deep sea – are ‘hit’ by faint light. Each time a sensor is hit, the position of that sensor in the sea and the time it was hit is recorded. The plots show on the y-axis the height of the sensors in the detector and on the x-axis the time. The red circles and the red line show how the light cone generated by a charged particle from below has crossed the detector. As function of time (in nanoseconds), the position of the next hit sensor is higher in the detector, indicating that the particle is travelling upwards. The blue circles are background hits.
With the installation of two more detection units at the French site of KM3NeT, the first phase of building the ORCA detector is completed! Since 27 January 2020, the detector is taking data with six detection units.
During a sea operation the 24-26 January 2020, the two new detection units have been connected to the KM3NeT/ORCA seafloor network at the KM3NeT/ORCA deep sea site, 40 km offshore from Toulon, France. The detection units were successfully positioned twenty metres apart.to within a metre of their target position 2.5 km below the sea surface. This highlights the skills of the staff on board the deployment ship, the precision of the custom acoustic positioning system and the maturity of the deployment method based on an innovative launching vehicle.
Immediate data taking
Using a robot, remotely operated from a second ship, the deployed units were connected to the seafloor network of the ORCA site. After a visual inspection of the detection units by the robot, the power was switched on and data taking with ORCA6 started immediately.
The ORCA detector has now six detection units – hence ORCA6. These are six vertical lines each with 18 sensor modules. A module houses 31 light sensors (photo-multiplier tubes) to record the faint Cherenkov light generated by charged particles in the sea water. That makes now 6x18x31= 3348 photo-multipliers in total in ORCA6. Each photo-multiplier records the intensity of the light flash and when it arrives. A compass, tilt meter and acoustic receiver record the position of the module in the sea water. With these measurements the path the charged particle took through the detector is precisely reconstructed.
The video below shows a selection of down-going cosmic rays (muons) passing through the ORCA6 detector soon after power up.
Important
Operating six detection units is an important milestone for KM3NeT as it marks the completion of the so-called ‘Phase 1’ of the project. In the next phase of KM3NeT/ORCA, the detector will be extended to 115 detection units.
After the general introduction by the KM3NeT Spokesperson Mauro Taiuti, the Deputy Spokesperson Aart Heijboer will present the expected performances of KM3NeT detectors and Dorothea Samtleben will show the first data from KM3NeT.
12 October 2020 – The KM3NeT Collaboration has published a new paper that aims at demonstrating the general applicability of deep convolutional neural networks to neutrino telescopes, using simulated datasets for the KM3NeT/ORCA detector as an example. To this end, the networks are employed to achieve reconstruction and classification tasks that constitute an alternative to the analysis pipeline presented for KM3NeT/ORCA in the KM3NeT Letter of Intent. They are used to infer event reconstruction estimates for the energy, the direction, and the interaction point of incident neutrinos. The spatial distribution of Cherenkov light generated by charged particles induced in neutrino interactions is classified as shower- or track-like, and the main background processes associated with the detection of atmospheric neutrinos are recognized. Performance comparisons to machine-learning classification and maximum-likelihood reconstruction algorithms previously developed for KM3NeT/ORCA are provided.
