Made up of the photons created in the Big Bang, the Cosmic Microwave Background (CMB) is the oldest light in the Universe. Tiny fluctuations in the temperature and polarization of the CMB across the sky encode the composition and history of the Universe. Ever since its accidental discovery in 1964, scientists have been detecting and decoding these signals with increasing sensitivity using telescopes on satellites, sounding rockets, balloons, and on the ground.
Most of these experiments conducted by research collaborations around the world had been separate efforts, but in 2013, at the Snowmass community planning meeting for particle physics and cosmology, the CMB community decided that to make significant progress, they needed to pool the resources of the national laboratories and universities into a collaborative effort to create the most sensitive instrumentation yet. Over the next two years, the DOE High Energy Physics Advisory Committee’s Particle Physics Project Prioritization Panel (P5) Report in 2014 and the National Academies of Sciences’ “2015 Strategic Vision for Antarctic and Southern Ocean Research,” recommended that DOE and the NSF Physics and Polar Programs support the development of a next-generation ground-based instrumentation and computational platform for cosmic microwave background detection.
A concept for this project was developed by an Astronomy and Astrophysics Advisory Committee task force in 2018. The following year, the DOE issued a “Mission Need” to work with the NSF on this effort—a milestone because it engaged the DOE national laboratory network’s significant fabrication, high performance computing, and project management resources in CMB science. Berkeley Lab was selected in the fall of 2020 to lead this multi-institution effort, named Cosmic Microwave Background Stage 4, or CMB-S4. The goal is to build 21 millimeter-wave telescopes at the South Pole and in Chile incorporating more than 550,000 highly sensitive sub-Kelvin, superconducting detectors, increasing sensitivity by over an order of magnitude compared with past projects.
Berkeley Lab Leads CMB-S4
The Lab’s lead role was due at least in part to its successes with other projects, including Dark Energy Spectroscopic Instrument (DESI), which recently won the DOE Project Management Excellence Award of Excellence, and the LUX-ZEPLIN project, a successful multi-institution underground detector project to detect dark matter.
“We have a good track record in managing projects, and in working with partner labs and the university community,” said Natalie Roe, associate Lab director for the Physical Sciences Area.
Today the CMB-S4 science collaboration numbers 307 members at 109 institutions in 18 countries and 28 U.S. states, including Berkeley Lab, Argonne National Laboratory, Fermi National Accelerator Laboratory, and SLAC National Accelerator Laboratory. It is the largest CMB collaboration to date, bringing together multiple sites and multiple sizes of telescopes to cover the entire range of CMB science. The University of Chicago leads the National Science Foundation’s part of the project, and works closely with Berkeley Lab to drive the effort.
At Berkeley Lab, the CMB-S4 team includes researchers from the Physical Sciences Area’s Physics Division, Engineering Division, from the Laboratory’s Project Management Office, and from the Computing Sciences Area’s Computational Research Division, National Energy Research Scientific Computing Center (NERSC), and Energy Sciences Network (ESnet).
In addition to hosting the Project Office for CMB-S4, Berkeley Lab is playing significant technical roles in the design and fabrication of the small aperture telescope arrays, of the superconducting cryogenic detectors, and in the data management systems. Detector fabrication is particularly challenging for CMB-S4, with the need to scale up an order of magnitude beyond prior efforts to fabricate these challenging devices. Berkeley Lab has successfully pioneered technology transfer to industry in order to increase the fabrication throughput of the superconducting detector wafers.
NERSC, ESnet, and the Computational Cosmology Center at Berkeley Lab will manage the very significant data needs of the project, providing high performance computing, networking, and software resources respectively as well as coordinating resources from other research institutions. Julian Borrill, who has been the Principal Investigator for NERSC’s CMB work since 1997 (pioneering the use of supercomputers to analyze CMB data shortly after NERSC moved to Berkeley Lab) and who led the U.S. data management effort for the space-based Planck project at NERSC from 2000 to 2015, is now leading the data management for the CMB-S4 ground-based project.
Said Julian, “Because we are searching for fainter signals, we have to collect larger sets of data. The computational resources thus have to manage an unprecedented volume of CMB data, and the data issues are complex. Our algorithms need to filter out the statistical noise, as well as account for inevitable imperfections in the experiment, such as slightly distorted beams or interference from ground signals. An additional challenge is that some of the newer supercomputers have sophisticated architectures that optimize computing power, but they are harder to program.”
The instrumentation, data collection, and data analysis effort, spanning a large number of institutions, is the most ambitious CMB project to date, one that aligns well with the Lab’s strengths. “CMB-S4 is a great example of what Berkeley Lab does so well – big science through multidisciplinary, multi-team efforts,” said Natalie.
Off to a Running Start
Already, the CMB-S4 project has gotten off to a strong start. The Decadal Survey on Astronomy and Astrophysics 2020 (Astro2020) released in November 2021 provided a strong endorsement of CMB-S4 as one of three large ground-based projects and the only one that is “shovel-ready.”
In addition, a four-day Director’s Review conducted by the DOE in November 2021, which aimed to assess the conceptual design, technical progress, and quality of the cost and schedule estimates developed by the project team, congratulated the project on the impressive progress made in the past year, especially in light of the limited funds available in the first year of the project. The review team noted that the science-driven technical design has advanced to or beyond the conceptual stage for most of the project and that the management structure, documentation, and planning is also very advanced for this stage. The reviewers also noted that research and development has progressed well to support the design and understanding and mitigation of technical risks, and that the management team is strong and experienced.
Said Natalie, “I’m very proud of what Berkeley Lab and the CMB-S4 team has accomplished. It’s gratifying that the DOE review committee recognized the progress that the team has made despite the constraints they faced.”
Integrating the Teams and Communicating during COVID
Multi-institution collaborations such as CMB-S4 require a lot of integration effort and infrastructure management. Interim project director John Corlett noted that there is one top level project management office for the DOE-NSF funded project, comprising representatives from the key institutions in the project. Planning is integrated. Operations, including finance, project controls, risk management, and procurement are all managed as a single office, resulting in greater efficiency and cost savings.
CMB-S4 has a large set of stakeholders to communicate with. In addition to the DOE, NSF, and project members, there are overseas partners that the CMB-S4 team would like to engage, and non-government funders like philanthropies that are helping support precursor experiments like CCAT-Prime Telescope Project and the Simons Observatory in Chile.
Not being able to travel or meet in person with stakeholders during the COVID pandemic has made communications and relationship-building more challenging. Since the Lab was assigned the lead role in August of 2020, the team has not held any in-person meetings.
John notes that the whole community has learned how to make it work. “We are working to keep everyone in the loop – scientists and engineers, as well as their managers and directors. There are lots of Zoom meetings, but we try to keep Zoom sessions shorter, with time for people to stretch and take breaks. We try hard to ensure our schedules and meetings are human-oriented,” he added.
The project members are now preparing for Critical Decision 1 review in late 2022, when the DOE Office of Project Assessment will examine a completed conceptual design submitted by the CMB-S4 team, analyze alternative approaches, and define a cost range for the project.
There is much to look forward to, according to Natalie. “There are lots of interesting opportunities for young and experienced researchers; CMB-S4 leverages cutting-edge hardware and computational expertise and will have an exciting and rich scientific program.”