Global End-to-End Analysis at All Frequencies

31 Aug 2026, 09:00 → 18 Sept 2026, 19:00 Europe/Stockholm

Albano Building 3

Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

 

Venue

Nordita, Stockholm, Sweden

Albano Building 3

Hannes Alfvéns väg 12, 10691 Stockholm, Sweden

 Go to map

---

Scope

Current and future astronomical experiments face unprecedented challenges in their calibration, systematics and modelling, as well as the interplay between them. The Cosmoglobe initiative aims to tackle these challenges head on, using the power of Bayesian end-to-end iterative analysis. By combining datasets from the radio, millimeter, infrared, visible, and all the way to the UV and X-ray ranges of the electromagnetic spectrum, we can jointly constrain cosmological parameters such as the tensor-to-scalar ratio, the optical depth to reionization, cosmic birefringence, the Cosmic Infrared Background and the structure of emissions from our own Galaxy. The meeting will focus on bringing together researchers working across modelling, simulation, and data analysis at all levels. It aims to foster discussion and collaboration within the community.

---

Themes and preliminary program schedule

NEWS: A detailed schedule for week two has been released here.

-------

Week 1 | Aug 31 - Sep 4 | CMB anisotropies workshop

• Tensor-to-scalar ratio /  The microwave sky / Systematics mitigation 
• E.g. HFI, LiteBIRD, SO, and related experiments
• Around 2 hours of presentations per day with lots of time allocated for discussions and active work sessions
• Lunch and coffee will be provided

 

Week 2 | Sep 7 - Sep 11 | Cosmoglobe conference 2026 

[Detailed schedule week 2.]

The topic of the conference is large scale science and data analysis across the entire spectrum. We therefore welcome and encourage conference contributions from anyone whose work fits into this broad category, regardless of whether you are already working directly with the Cosmoglobe team.

• Global end-to-end analysis at all frequencies from radio to UV and X-ray
• E.g. SKA and precursors, HFI, LiteBIRD, SO, C-BASS, QUIJOTE, DIRBE, AKARI, FIRAS, IRAS, Fermi, eROSITA, etc. 
• Submitted and invited talks on all things related to global sky modelling 
• Lunch and coffee and a conference dinner (tentatively Sept. 8th) will be provided
• We welcome contributions from early career researchers!

 

Week 3 | Sep 14 - Sep 18 | Global sky modelling workshop 

• Infrared dust modelling / CMB spectral distortions 
• E.g. DIRBE, AKARI, FIRAS, IRAS, and related experiments
• Around 2 hours of presentations per day with lots of time allocated for discussions and active work sessions
• Lunch and coffee will be provided

 

---

Speakers

The current list of confirmed participants can be found here. 

NEWS: The speaker list can be found here.

---

Accommodation / Housing support

[Registration closed] There is limited housing available through Biz Apartments, which will be given preferentially to early career researchers who would not be able to attend otherwise. Please indicate in the registration form if you would like to be considered for housing support or whether you will arrange your own accommodation.

It is also possible to book rooms yourself directly via Biz Apartments website if you prefer to stay at the conference housing. Currently they have 15% discount for online bookings. Note, that there are other hotels closer to Nordita as well. 

---

Travel support

[Registration closed] There is a limited amount of travel support available, which will be given preferentially to early career researchers who would not be able to attend otherwise. If this applies to you, please fill out that section of the registration form.

---

Application/Registration

[Registration closed] Registration to be considered for on-site participation will close May 17! Registrants will receive a participation confirmation from the organizers after this date.

Note, you will receive an email with confirmation of completion of the application form after submission. If you don't receive this email please try again from a different browser or contact Nordita via event@nordita.org. 

---

VISA / Certificate of Attendance

To check whether you need a VISA please access the Schengen Visa Info site as soon as possible, note that VISAs can take >1 month to obtain. 

If you require an invitation letter for the conference/workshop, please send an email to cosmoglobe-contact [at] astro.uio.no with your

• Name, surname, and date of birth
• Passport number, country of issue, and expiration date

 

---

Sponsored by:

OpenHFI  - RCN Young Research Talents agreement no. 360024 - PI M. Galloway - 2026-2030

LiteBIRD-Norway - RCN FRIPRO agreement No. 351037 – PI U. Fuskeland - 2025-2033

Origins - EU ERC-StG agreement No. 101165647 – PI D. Watts - 2025-2030

Commander - EU ERC-AdG agreement No. 101141621 – PI H. K. Eriksen - 2024-2029

CosmoglobeHD - RCN Young Research Talents agreement no. 344934 – PI D. Watts - 2024-2027

Monday 7 September

09:00 → 09:45 Coffee and registration

09:45 → 10:00 Welcome

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

10:00 → 10:25 Cosmoglobe - mapping the sky from the Milky Way to the Big Bang

Speaker: Ingunn Kathrine Wehus (University of Oslo)

10:25 → 10:45 Cosmoglobe DIRBE DR2 results

In this talk, I will give an overview of the Cosmoglobe DR2 analysis, in particular the application of global end-to-end processing to COBE/DIRBE data. As an absolutely calibrated instrument, DIRBE is uniquely sensitive to the cosmic infrared background monopole from 1 micron to 240 microns. I will give an overview of the improved time-ordered data processing that results in maps with significantly reduced zodiacal dust contamination, as well as improved CIB monopole values consistent with theoretical predictions from integrated galactic light models.

Speaker: Duncan Watts

10:45 → 11:00 Zodiacal light modelling in Cosmoglobe DR2

I will present an improved zodiacal light (ZL) model for COBE-DIRBE derived through global Bayesian analysis within the Cosmoglobe Data Release 2 framework. The parametric form of the ZL model is inspired by the original DIRBE model by Kelsall et al. (K98), but the specific best-fit parameter values are re-derived using the combination of DIRBE Calibrated Individual Observations, Planck HFI sky maps, and WISE and Gaia compact object catalogs.

Speaker: Angela Bonato (University of Oslo)

11:00 → 11:30 Coffee

11:30 → 11:55 The Simons Observatory: Early Data from the Small Aperture Telescopes

The Simons Observatory (SO) is a cosmic microwave background survey experiment located in the Atacama Desert in Chile. SO consists of multiple small-aperture telescopes (SATs) carrying out a focused small-area survey, as well as a large-aperture telescope (LAT) conducting a wide-field, high-resolution survey. The SATs are specifically designed to target primordial B-mode polarization sourced by tensor perturbations in the early Universe. The SO mid-frequency SATs saw first light in late 2023, and have since been conducting initial science observations. I will present the current status of and early data from the SATs and the SAT data processing pipeline.

Speaker: Erik Rosenberg (University of Manchester)

11:55 → 12:20 Polarization from Planck 857GHz data

Planck survey may have completed in 2013 but the data analysis continues. I will give an overview of a recent analysis of the 857GHz channel making use of the slight residual polarization sensitivity of the nominally unpolarized spiderweb bolometers.

Speaker: Reijo Keskitalo (University of Oslo)

12:20 → 12:45 New State-of-the-Art Constraints on Cosmic Birefringence from OpenHFI Maps

Cosmic birefringence ($\text{CB}$)—the in-vacuum rotation of the linear polarization plane of photons—presents a compelling hint of parity-violating physics beyond the Standard Model. Recent analyses have brought this effect to the forefront of $\text{CMB}$ cosmology: Planck PR4 data yielded an estimate of $\beta = 0.3^\circ \pm 0.05^\circ \text{ (stat)} \pm 0.28^\circ \text{ (sys)}$, while the Atacama Cosmology Telescope ($\text{ACT}$) DR6 independently found $\beta = 0.2^\circ \pm 0.07^\circ$. Together, their combined significance stands at $4.6\sigma$, hovering just below the definitive discovery threshold.The primary bottleneck in confirming or ruling out this signal lies in the meticulous control of instrumental systematic effects, particularly polarization angle calibrations. In this talk, we discuss how to leverage the OpenHFI end-to-end $\text{CMB}$ maps, which benefit from the comprehensive systematic mitigation framework of the Cosmoglobe endeavor. By applying different $TB$ and $EB$ power spectra-based estimators to these freshly processed Planck HFI products, we demonstrate how to tightly constrain the cosmic birefringence angle. Finally, we show how this enhanced control over systematic contaminants impacts the total error budget and explore the potential of these maps to push the current detection significance past the $5\sigma$ threshold.

Speaker: Simone Paradiso (INAF-OAS)

12:45 → 13:00 Avenues for improving dust foreground modeling with stellar polarization

The alignment of dust grains with the Galactic magnetic field gives rise to polarized dust emission and polarization of starlight in the optical/NIR. This results in a tight correlation between the polarization angles of distant stars and that of dust emission. I will give an overview of recent works that use stellar polarimetry and stellar distances to inform foreground modeling, including disentangling different emitting regions along the line of sight, and quantifying the effects of frequency decorrelation. I will discuss how the tight correlation between the Stokes parameters of dust emission and starlight polarization found by Planck can be used to construct higher angular resolution polarized emission templates than are presently available. Finallly, I will present new results that suggest that the optical-FIR polarization correlation varies over the sky and discuss implications for foreground models and for deducing the microphysical properties of dust in the diffuse ISM.

Speaker: Gina Panopoulou (Chalmers University of Technology)

13:00 → 14:30 Lunch

14:30 → 14:45 Constraining Low-Frequency Foregrounds with CHIPASS

One of the greatest challenges faced by CMB experiments is the detailed modeling and subtraction of Galactic foregrounds. Observations across multiple frequencies are necessary to constrain the synchrotron SED as well as to disentangle it from other foregrounds such as free-free emission. CHIPASS, a continuum reprocessing of the HI Parkes All Sky Survey (HIPASS) with an angular resolution of 14.4 arcmin, offers a unique window on the synchrotron foreground with the potential to constrain spectral index variations and curvature as well as to reveal small-scale structures. The Cosmoglobe collaboration has begun integrating in raw time-ordered data from CHIPASS in combination with other low-frequency datasets such as Haslam, S-PASS, and others. Here we present a preliminary look at the re-analyzed CHIPASS data.

Speaker: Danielle Sponseller (Chalmers University of Technology)

14:45 → 15:10 All-Sky Modelling of Galactic Emission at Radio-Microwave Frequencies

I will begin by introducing the C-BASS survey and its role as a key low-frequency radio dataset for Galactic foreground studies, before presenting a new all-sky model of low-frequency diffuse Galactic emission describing the regime where synchrotron, free-free, and spinning dust emission dominate. The model extends the Planck 2015 diffuse component-separation by incorporating more recent radio and microwave surveys. We fit 35 full- and partial-sky maps, at 1 degree resolution, including S-PASS (2.30 GHz), C-BASS (4.76 GHz), and QUIJOTE (10-20 GHz), together with reprocessed WMAP and Planck LFI data from the Cosmoglobe collaboration and Planck HFI channels. Using a Bayesian parametric approach with Commander, we derive spatially varying amplitude and spectral parameter maps for the dominant low-frequency foreground components in total intensity. The main products of this work are: (i) a new full-sky synchrotron amplitude and spectral index solution, (ii) an all-sky characterisation of spinning dust emission based on a single-component log-normal spectral model, including peak frequency and width, and (iii) a new all-sky reconstruction at 4.76 GHz providing a tracer of diffuse synchrotron emission with reduced systematics relative to the Haslam map (408 MHz). These products describe the transition between radio and microwave emission and provide a new reference for foreground modelling and sky-simulation applications.

Speaker: Gabriel Hoerning (University of Manchester)

15:10 → 15:35 Diffuse Polarised Synchrotron Reconstruction with GNILC On Latest Planck Data

Primordial gravitational waves from inflation imprint a faint B-mode polarisation in the Cosmic Microwave Background (CMB), but this signal is strongly obscured by Galactic foregrounds. Detecting CMB B-modes therefore requires accurate characterisation of foreground polarisation, dominated by synchrotron emission at low frequencies. To address this, we apply the Generalised Needlet Internal Linear Combination (GNILC) method to Planck PR4 data (30–353 GHz), producing clean, low-noise synchrotron polarisation maps at low frequencies without relying on prior models. GNILC employs needlets for localised processing in both pixel and harmonic space, a principal component analysis to retain modes above noise, and a multi-frequency weighting scheme orthogonal to the CMB spectrum to deproject the CMB. Radio source inpainting prior to GNILC processing further reduces contamination in the resulting maps. The method is validated on Planck NPIPE simulations, and PR4-derived GNILC weights are applied to NPIPE noise, systematics and CMB realizations to provide robust residual error characterisation of the GNILC PR4 maps. The final GNILC PR4 maps at 30 and 44 GHz provide reliable full-sky diffuse polarised synchrotron templates at uniform 34' angular resolution, as demonstrated by map inspection, residual error estimates, power spectra, and low correlation with Planck cleaned CMB maps. Exploiting the frequency scaling between our 30 and 44 GHz GNILC PR4 maps, we derive a first-of-its-kind 48-pixel full sky synchrotron spectral index map from Planck-only data, with the associated uncertainty.

Speaker: Mohammad Ishaque Khan (Instituto de Fisica de Cantabria (IFCA, CSIC-UC))

15:35 → 16:00 The Canadian Galactic Emission Mapper: Overview and Early Observations

Gravitational waves from inflation may have left a detectable signature in the parity-odd, “B-mode” component of the polarization of the Cosmic Microwave Background (CMB). Detecting B-modes in the CMB would be “smoking gun” evidence for inflation and would probe some of the highest-energy physics in the known universe. However, current experiments have placed stringent upper limits on B-modes. If B-modes are present in the CMB, the signal is extremely faint and is dominated by polarized Galactic foregrounds at all frequencies. It is therefore essential to map polarized foregrounds with high precision to enable a detection of CMB B-modes. The Canadian Galactic Emission Mapper (CGEM) is a new 4m single-dish radio telescope at the Dominion Radio Astrophysical Observatory that is mapping polarized Galactic synchrotron emission from 8-10GHz over the Northern sky, with a planned counterpart in the Southern hemisphere. CGEM will greatly improve models of polarized CMB foregrounds and will hence be an important aid to current and future B-mode experiments. In this talk, I’ll give an overview of CGEM. I’ll then describe how we designed this purpose-built instrument to measure the sky with minimal polarization systematics, highlighting in particular the polarization purity of the optical design. I’ll also showcase early observations from a pathfinder version of CGEM, deployed in February 2025, which has been mapping the sky since. These data already show immense promise for future science with CGEM.

Speaker: Joshua MacEachern (National Research Council of Canada, Dominion Radio Astrophysical Observatory)

16:00 → 16:30 Coffee

16:30 → 16:45 Non-Gaussian statistics and machine learning for cosmological inference with 21cm intensity mapping

Cosmology is currently experiencing a data-rich era, with even more data expected in the coming decade. Intensity mapping of 21 cm neutral hydrogen emission offers a promising avenue for mapping vast volumes of the Universe with high redshift resolution, aiding in the study of large-scale structure and cosmological parameter estimation. However, extracting cosmological information from these datasets poses challenges, including significant foreground contamination, instrumental effects, and the loss of information inherent in using two-point statistics in highly non-Gaussian fields. This ongoing PhD project aims to explore the potential of higher-order statistics—such as Minkowski functionals, peak counts, and probability distribution functions—combined with machine learning techniques for cosmological inference from 21 cm data. The research is organized into three main topics: (i) assessing the robustness of these statistics against foreground residuals and instrumental imperfections; (ii) quantifying their cosmological information content using N-body simulations through a simulation-based inference approach (SBI); and (iii) developing an emulator for higher-order statistics and comparing different cosmological inference methodologies. The results and tools generated from this project will significantly contribute to the scientific preparation for future 21 cm experiments, such as SKA, the largest experiment of its kind currently under construction, and BINGO, a radio telescope being developed in Brazil.

Speaker: Gabriel Silva Costa (National Institute for Space Research (INPE))

16:45 → 17:10 From Few Simulations to Trustworthy Posteriors: Scattering-Covariance Latent Spaces for Simulation Augmentation and Interpretable SBI in CMB Cosmology

Modern CMB analyses face a statistical bottleneck: the data are high-dimensional, non-Gaussian, affected by complex systematics, and often have intractable likelihoods, while the end-to-end simulations needed for Monte Carlo validation or Simulation-Based Inference are prohibitively expensive. I will present a unified framework based on the Scattering Covariance: an interpretable, physics-informed analogue of a convolutional neural network, built from fixed oriented wavelets, nonlinearities, and cross-scale/cross-channel covariance statistics. In recent work ([Campeti et al., A&A, 2025][1]), we used this representation to construct a fast map-level generative emulator for CMB instrumental systematics simulations. Even when trained on as few as ten high-fidelity simulations, the emulator generates statistically independent approximate realizations that reproduce power spectra, scattering statistics, Minkowski functionals, and pixel-covariance structure, enabling orders-of-magnitude simulation augmentation at negligible cost compared with full end-to-end campaigns. I will then describe how the same scattering-covariance latent space can be used for transparent SBI pipelines for CMB polarization in Planck, LiteBIRD and Simons Observatory data, to infer parameters such as the optical depth to reionization and the tensor-to-scalar ratio while controlling non-stationary foreground and instrumental residuals. The goal is a simulation-efficient, interpretable alternative to ``black-box'' neural inference: using a few expensive simulations as anchors for large, calibrated, statistically-controlled inference pipelines. [1]: https://www.aanda.org/articles/aa/full_html/2025/08/aa54540-25/aa54540-25.html

Speaker: Paolo Campetti (University of Ferrara)

17:10 → 17:15 Summary of the day

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

17:30 → 19:00 Welcome reception

Tuesday 8 September

09:30 → 10:00 Coffee

10:00 → 10:05 Welcome

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

10:05 → 10:35 Three decades of cosmological Gibbs sampling

I will give a historical overview of how the method of Gibbs sampling was developed and applied during (soon) almost three decades of continuous work by a large group of people.

Speaker: Hans Kristian Eriksen (University of Oslo)

10:35 → 11:00 The FOSSIL mission

Speaker: Nabila Aghanim (IAS)

11:00 → 11:30 Coffee

11:30 → 11:45 AKARI and the infrared sky

AKARI’s Far-Infrared Surveyor (FIS) produced high-angular-resolution all-sky observations in four bands spanning 65-160 μm, offering a valuable view of thermal emission from interstellar dust, zodiacal light, and the cosmic infrared background. This talk provides an overview of the instrument, survey, and its place among infrared and microwave observations, emphasising synergies with current and upcoming surveys and foreground modelling relevant for CMB analyses. I will close by presenting preliminary results from our ongoing AKARI reanalysis within Cosmoglobe.

Speaker: Katrine Alice Glasscock (University of Oslo)

11:45 → 12:10 Cosmoglobe re-analysis of AKARI-FIS

In this talk, I will introduce the re-analysis of AKARI-FIS currently underway within the Cosmoglobe collaboration. I will discuss the potential of the AKARI observations to shed light on the Cosmoglobe sky model, in particular with regards to interstellar and zodiacal dust modelling as well as observations of the CIB. I will also showcase some of the AKARI time-ordered data, its characteristics and features, and the various ways they were tackled by the original AKARI team, and the ways in which we aim to tackle them within the Cosmoglobe framework, which has the benefit of being able to use other data sets and an existing sky model to shed light on the various systematics within AKARI, while also being able to optimally extract new science from the AKARI data itself to update said sky model.

Speaker: Eirik Gjerløw (University of Oslo)

12:10 → 12:35 Mitigating correlated noise in map-making for CMB experiments with BrahMap

Low-frequency correlated noise is a critical systematic for next-generation CMB experiments targeting the primordial B-mode signal. 1/f noise residuals are known to contaminate large angular scales and bias the estimation of the tensor-to-scalar ratio if they are not mitigated properly. In this context, we present BrahMap, a scalable and modular map-making framework designed for large-scale data reduction for current and future CMB polarization experiments. It features a user-friendly Python interface backed by performance-optimized C++ extensions with MPI+OpenMP parallelization. BrahMap implements generalized least squares (GLS) map-making with a large suite of noise covariance operators - from diagonal and circulant approximations to full Toeplitz operators - all accessible through a unified linear operator algebra interface. We explore several strategies for mitigating 1/f noise at the map-making level. First, we perform a systematic comparison of different circulant approximations of the exact noise covariance and assess their noise mitigation performance. Next, we implement superfast Toeplitz solvers to enable computing the exact GLS estimate using the full Toeplitz noise covariance. With realistic LiteBIRD simulations across a wide range of 1/f noise regimes, we demonstrate that the exact Toeplitz approach reduces the large-scale noise residual power by up to two orders of magnitude compared to circulant approximations, translating to an approximately four-fold reduction in the bias on the tensor-to-scalar ratio.

Speaker: Avinash Anand (University of Rome "Tor Vergata")

12:35 → 12:50 CG mapmaking for AKARI in Commander4

The most straightforward way to construct a map from a set of time-ordered data (TOD) is through a classic inverse-variance bin mapmaker. However, certain experiments such as AKARI and Planck HFI require to take into account non-local effects, such as the bolometer transfer function or detector cross-talks. In such case, inverting the mapmaking equation analytically is not feasible and a conjugate gradient search is required for retrieving the map solution, i.e. CG mapmaking.

This talk aims to give an insight on the implementation of CG mapmaking in Commander4, in relation to AKARI TOD reanalysis.

Speaker: Leo Bianchi (University of Oslo)

12:50 → 14:20 Lunch

14:20 → 14:45 Challenges of 21cm Radio Data Analysis

The 21cm signal from neutral hydrogen traces cosmic large-scale structure in 3D over an exceptionally wide redshift range. It has proven difficult to disentangle from foreground contamination and instrumental effects however. I will explain how the large dynamic range between signal and foregrounds and stringent spectral calibration requirements make this a difficult data analysis problem. I will then review the Hydra Gibbs sampling framework, which can handle low-frequency radio data from different types of radio telescope.

Speaker: Phil Bull (University of Manchester)

14:45 → 15:00 Making Absolutely Calibrated Measurements with Gaussian Constrained Realisations for the RHINO Global 21-cm Signal Experiment

The Remote HI eNvironment Observer (RHINO), currently under construction at Jodrell Bank Observatory, is an experiment aiming to observe between 60-85 MHz to measure the sky-averaged 21-cm absorption feature redshifted from the Cosmic Dawn. This signal is obscured by galactic synchrotron foregrounds that are brighter by over four orders of magnitude. Additionally, instrumental systematics impart additional spectral structure that can hinder extraction of the cosmological signal if not sufficiently calibrated for. This talk will outline how RHINO employs a Bayesian approach to calibrating instrumental systematics using Gaussian Constrained Realisations and propagating uncertainties into absolutely calibrated measurements of the low-frequency sky.

Speaker: Jordan Norris (University of Manchester)

15:00 → 15:15 Beam Systematics and Absolute Calibration in Global 21 cm Measurements

Global 21 cm experiments measure the absolute sky temperature at low radio frequencies, offering a potential low-resolution reference for large-scale sky models. However, the measurement is strongly shaped by the antenna beam, whose imperfect calibration can mix spurious sky structure into the observed spectrum. I present simulations testing whether controlled antenna rotation can help separate beam effects from true sky signals. By comparing different rotation strategies with realistic beams and sky models, this work explores a practical route toward improved absolute calibration for global sky analysis.

Speaker: Rashi Srivastava (University of Manchester)

15:15 → 15:40 Observations of the full northern sky at 10-40 GHz with the QUIJOTE experiment

In this talk I will give an overview of the status and recent results of QUIJOTE, a CMB polarization experiment working in the frequency range 10-40 GHz from the Teide Observatory (Tenerife, Spain). The QUIJOTE MFI instrument produced maps covering the full northern hemisphere at 11, 13, 17 and 19 GHz that were publicly released in 2023. I will highlight the scientific results coming out of these maps, addressing the measurement of the synchrotron spectrum in polarisation, the determination of the level of polarisation of the AME, and the study of the correlations of the AME with dust emission and with different environmental parameters. I will also briefly comment on the status of new QUIJOTE instrumentation at 10-20, 30 and 40 GHz, and of other CMB experiments operative (or to become operative) from the Teide Observatory (GroundBIRD, LSPE/Strip, TMS).

Speaker: Ricardo T. Génova-Santos (Instituto de Astrofísica de Canarias)

15:40 → 16:00 Conference picture!

16:00 → 16:30 Coffee

16:30 → 16:55 Probing the AME and PAH Connection via Multi-Frequency PCA

Anomalous Microwave Emission (AME) remains a critical uncertainty in global sky modeling, with its widely hypothesized origin tied to spinning Polycyclic Aromatic Hydrocarbons (PAHs). Robustly separating AME from other diffuse Galactic foregrounds requires using multi-frequency data across the electromagnetic spectrum. In this work, we present a blind component separation approach using Principal Component Analysis (PCA) applied to a comprehensive dataset spanning radio to infrared wavelengths. Our dataset synthesizes observations from space-based missions (Planck, WMAP, DIRBE, Haslam) and incorporates critical ground-based surveys (e.g., C-BASS, QUIJOTE). To physically interpret the extracted PCA components, we introduce a novel framework bridging blind and parametric techniques. We construct a theoretical basis via a moment expansion of standard foreground Spectral Energy Distributions (SEDs). By orthonormalizing this model basis, we can predict the uncorrelated spatial maps expected from the underlying physics. If the parametric model accurately describes the sky, its basis will span the exact same subspace as the data-derived PCA eigenvectors. We have successfully validated this methodology on simulated skies, demonstrating a high spatial correlation between the model-predicted components and the blind PCA maps. Building on this success, we will present the application of our pipeline to observational datasets, with the ultimate goal of isolating the AME component and robustly testing its correlation with infrared PAH tracers.

Speaker: Elena de la Hoz (NASA Goddard Space Flight Center)

16:55 → 17:10 Small-scale Non-Gaussianity of Galactic Foreground Emissions

Small-scale non-Gaussianity in Galactic foregrounds is an important source of uncertainty for future CMB experiments. We investigate the statistical properties of diffuse Galactic foregrounds using Minkowski Functionals and generalized skewness–kurtosis statistics across a range of angular scales. We find a remarkably similar kurtosis-type non-Gaussian signature in synchrotron, free-free, anomalous microwave emission, and thermal dust. Focusing on thermal dust, we compare Planck GNILC and 545 GHz observations with PySM and DUSTFILAMENTS simulations and perform patch-based analyses to study the spatial distribution of the signal. While PySM reproduces the global statistical trends, it fails to capture the observed spatial variations, and DUSTFILAMENTS underestimates the measured non-Gaussianity. Using controlled toy models with different one-point PDFs and power spectra, we show that the observed behaviour naturally arises from the combination of a heavy-tailed one-point PDF and a steep dust-like power spectrum. Finally, I will discuss our ongoing efforts to improve dust simulations by incorporating observed filament structural and spectral properties.

Speaker: Fazlu Rahman Panam Parambil (Texas A&M University)

17:10 → 17:15 Summary of the day

18:00 → 22:00 Conference dinner

Wednesday 9 September

09:30 → 10:00 Coffee

10:00 → 10:05 Welcome

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

10:05 → 10:20 High resolution dust analysis with DIRBE and AKARI

The Cosmoglobe collaboration has recently completed the second data release, the reanalysis of the COBE-DIRBE data, producing new starlight emission models, zodiacal light models, cosmic infrared background (CIB) detections and new Milky Way dust models. In this talk, I will focus on the new dust models. This new dust model simplifies previously used dust models, and improves the interpretation, by breaking the old single component dust model into four physically motivated dust components. This is further motivated by the fantastic fit the dust has to both the DIRBE and Planck HFI data. Dust represents a significant systematic noise contribution to cosmic microwave background observations, so this new dust model will be crucial for future detection of signals important for understanding the earliest moments of our Universe, cosmic inflation.

Speaker: Raelyn Sullivan (University of Oslo)

10:20 → 10:45 Improved foreground cleaning via polarization-aided cleaning and CIB deprojection in thermal SZ Compton-y maps

The thermal Sunyaev-Zeldovich effect Compton-y parameter map (y-map) is a redshift-independent tracer of the hot baryon density across the sky. The all-sky y-maps reconstructed from Planck data have been utilized for the study of Lambda-CDM cosmology and the astrophysics of galaxy clusters. The current best minimum variance, all-sky y-map was made public in our previous work, using Needlet ILC on the latest Planck data. I will describe our effort to develop the best strategies for reconstructing y-maps with the least possible contamination from galactic foregrounds using the Hybrid ILC approach, and with the least cosmic infrared background (CIB) contamination using optimal deprojection of moments of the CIB SED. I will also compare our CIB minimization with other contemporary works with the same aim.

Speaker: Jyothis Chandran (Instituto de Fisica de Cantabria (IFCA, CSIC-UC))

10:45 → 11:00 Untangling dust emission and CIB anisotropies with the Scattering Transform Statistics

Template-fit approach is often used to separate the Galactic dust emission and the cosmic infrared background (CIB) anisotropies at low HI column density regions with an underlying assumption that the gas and dust are tightly correlated. However, this method fails in regions where additional Galactic emission from the molecular hydrogen, diffuse ionized gas, and dark gas are present. We develop and test a statistical component separation to extract the dust signal from the contaminated Planck 353 GHz observations using the Scattering Transform (ST) statistics. We apply two sub-classes of ST statistics to regions with high and low signal-to-noise ratios, where the dust-HI correlation breaks down, and extract the dust emission from beneath the CIB anisotropies.

Speaker: Srijita Sinha (NISER Bhubaneswar, RRI, Bangalore)

11:00 → 11:45 Posters and coffee

11:45 → 12:10 Forecasted constraints on Reionization and Neutrino masses from LiteBIRD

LiteBIRD is an ISAS/JAXA strategic L-class mission to probe cosmic inflation and the cosmic history of the Universe through full-sky measurements of cosmic microwave background (CMB) polarization. Its design will allow an almost cosmic variance limited measurement of the E-mode power spectrum at large angular scales and a much tighter constraint on the optical depth of reionization $\tau$ compared to Planck. We forecast how LiteBIRD, in combination with Planck CMB data for the smaller angular scales, will improve the constraints on $\tau$ and the reionization history. We perform an MCMC analysis using two mock datasets with different reionization histories and assuming four different reionization models. We see that LiteBIRD can constrain $\sigma(\tau)$ approximately three times better than Planck, and constrain the parameters of more general reionization models. If the assumed reionization model does not fit well the mock data reionization history, the best-fit $\tau$ from LiteBIRD can be biased. Combining LiteBIRD with the Simons Observatory (SO) CMB small-scales, SO lensing and Baryon Acoustic Oscillations (BAO), we forecast how the tighter $\sigma(\tau)$ from LiteBIRD translates into an improved constraint on the sum of neutrino masses $\Sigma m_{\nu}$. We also see that assuming the wrong reionization model would bias not only $\tau$, but also $\Sigma m_{\nu}$.

Speaker: Serena Giardiello (Cardiff University)

12:10 → 12:25 Reionization Constraints from a combined Planck+WMAP Low-ℓ Analysis

Reionisation marks the epoch when the first luminous structures ionised the intergalactic medium after recombination, producing free electrons that Thomson-scatter CMB photons. This imprint is encoded in the large-scale E-mode polarisation of the CMB, which directly constrains the optical depth τ. Despite the high quality of Planck and WMAP data, a fully consistent framework that jointly exploits all low-ℓ polarisation information has been missing. To fill this gap, we present ELiCA (E-mode Likelihood for Cross-Analysis), an harmonic-based unified likelihood that consistently combines Planck (100 GHz and 143 GHz HFI channels and 70 GHz LFI channel) and WMAP (Ka, Q, and V bands) large-scale polarisation data. Validated on simulations, ELiCA delivers robust and precise constraints on the optical depth: τ = 0.0581^{+0.0048}_{-0.0059} (68% CL). This result strongly disfavors high-τ scenarios associated with early reionisation, reinforcing a low optical depth picture. The improved precision propagates to ΛCDM extensions, yielding a stringent upper bound on the summed neutrino mass: Σmν < 0.069 eV (95% CL). ELiCA sets a new benchmark for large-scale CMB polarisation analyses and provides a natural framework for next-generation missions such as LiteBIRD, where precision measurements of τ will be critical for reionisation physics and cosmological tensions. The ELiCA likelihood is publicly available.

Speaker: Valentina Genesini (University of Ferrara)

12:25 → 12:40 Exploring Rayleigh Scattering in the Era of Precision CMB Polarisation Measurements

Measurements of the Cosmic Microwave Background (CMB) temperature anisotropies over the past three decades have transformed our understanding of the Universe, while polarisation measurements have begun to provide additional powerful cosmological insights. Upcoming CMB experiments such as LiteBIRD, the Simons Observatory, and future missions including PICO will measure CMB polarisation with unprecedented sensitivity across a wide range of frequencies. In the standard cosmological framework, recombination physics is typically modelled considering only Thomson scattering. However, an additional and unavoidable process — Rayleigh scattering of CMB photons by neutral hydrogen and helium formed during recombination — becomes significant at high observational frequencies (≳200GHz). This effect introduces a characteristic frequency dependence in the CMB anisotropies and provides a new avenue for improving cosmological parameter estimation by breaking existing parameter degeneracies. In this talk, I will discuss the prospects for detecting the Rayleigh scattering signal with future CMB missions such as LiteBIRD and PICO, and quantify its impact on cosmological constraints. I will further highlight how Rayleigh scattering can serve as a sensitive probe of recombination physics and explore its role in testing alternative recombination scenarios aimed at alleviating the Hubble tension.

Speaker: Debabrata Adak (Institute of Astrophysics of the Canary Islands)

12:40 → 13:00 Absolute Gain Calibration for MeerKLASS HI Intensity Mapping: Point Sources, Noise Diodes, and RFI Mitigation

The MeerKLASS survey uses MeerKAT in single-dish autocorrelation mode to map the large-scale neutral-hydrogen sky over hundreds of square degrees, with the long-term goal of contributing a low-redshift, radio-frequency view of cosmic structure to the broader multi-wavelength sky-modelling effort. Reaching the required sensitivity demands gain calibration that is stable and absolute to a level well beyond what is needed for typical radio imaging, and that can be propagated cleanly into an end-to-end analysis alongside foreground and systematics models. In this poster I present recent progress within the MuSEEK pipeline on tying together the two principal calibration handles available to a single-dish experiment: bright astronomical point sources (Hydra A, Pictor A) observed at the start and end of each scan, and the injected noise diode firings that bracket every dump. Point-source calibration provides an absolute temperature anchor through a physical model combining the primary beam, atmospheric opacity, receiver temperature, and ground spillover, while the noise diode tracks the bandpass and short-timescale gain drifts between calibrator visits. I show the level of consistency between the two approaches across the MeerKAT UHF band (~580–1015 MHz), the residual spectral structure that limits their agreement, and the implications for the per-dish gain solutions propagated downstream into map-making. A persistent obstacle to both calibration paths is low-level RFI that survives conventional flagging and biases gain estimates in subtle, frequency-dependent ways — a systematic that directly leaks into any global model built on these data. As part of an ongoing effort within the MeerKLASS collaboration to harden the calibration chain, I am evaluating a number of recently developed tools against real UHF observations, including a moment-based RFI detection scheme (momentRFI) intended to complement AOFlagger by exploiting higher-order statistics of the visibility distribution to catch contamination that escapes amplitude-threshold flaggers. I report on its impact, and that of related improvements, on the recovered gain solutions and on the noise-diode-derived bandpass, and close with an outlook on folding the most promising of these into the standard MeerKLASS calibration chain and on how the resulting data products are intended to interface with end-to-end Bayesian analyses of the kind central to the Cosmoglobe programme.

Speaker: Geoff Murphy (University of Western Cape)

13:00 → 14:30 Lunch

14:30 → 14:55 The Simons Observatory: A framework for efficient map-based simulations of instrumental systematics for CMB surveys

The current and future observations of the CMB are expected to reach unprecedented sensitivity to place stringent constraints on both cosmological parameters and fundamental physics. The ground-based Simons Observatory Large Aperture Telescope started observations last year, and will provide the most sensitive small-scale CMB measurements. A tight control over instrumental systematics is critical, through the assessment of their impact as well as their mitigation. In this talk, we propose a new generalized framework to efficiently simulate systematics at the map level, along with the Python package smarties optimized to handle high resolution simulations. As a first application, we demonstrate that our results are in agreement with significantly more numerically expensive end-to-end simulations for a set of beam-related systematics. The ability to efficiently produce large ensembles of such simulations will allow for more robust cosmological parameter estimation from the power spectra of the CMB and gravitational lensing potential.

Speaker: Magdy Morshed (INFN, Ferrara)

14:55 → 15:20 Time-ordered and map-based simulations of beam systematics for CMB experiments

I will present work done with Magdy Morshed on a new, fast way of simulating the temperature to polarization leakage due to beam pointing errors and mismatched ellipticity in detector pairs at the map level, as well as how we validated it using the TOAST time-domain simulation pipeline. Those effects are important for instruments that aim to measure the small scales of the polarized CMB, like the SO LAT.

Speaker: Alexandre Adler (UC Berkeley/LBNL)

15:20 → 15:45 BayesLIM: End-to-end, differentiable, Bayesian forward models for joint 21cm signal, foreground, and instrumental parameter inference

Next-generation 21cm cosmology experiments promise to revolutionize our understanding of the high redshift universe; however, these experiments are overwhelmed by foregrounds many orders of magnitude brighter. This sets a precise specification on instrumental modeling requirements that have to date made direct detection of the 21cm at the EoR elusive. However, next-gen computing and inference paradigms such as auto-diff and GPU acceleration will make joint modeling of the 21cm signal, foregrounds, and systematics possible, even for large-scale interferometric datasets. I'll discuss the BayesLIM approach, the first end-to-end, differentiable Bayesian forward model for low-frequency LIM, and the path forward for cosmological end-to-end modeling at low frequencies. (Arxiv:2504.07090)

Speaker: Nicholas Kern (University of Michigan)

15:45 → 16:00 Mapping the Star-Forming Universe with the COMAP

Modern cosmology has revealed many secrets of the nature and evolution of the Universe. To date, most surveys have done this by mapping the early Universe with the Cosmic Microwave Background (CMB) or relatively recent times by mapping the large-scale distribution of galaxies. However, epochs in between, from the dark ages through the epoch of reionization (EoR) all the way to the Epoch of Galaxy Assembly (EoGA), are strongly data-starved. This is where line intensity mapping aims to fill the gap by providing direct 3D measurements of the aggregate emission from redshifted lines such as the 21 cm line, CO rotational lines, [CII], and other atomic and molecular lines. These will be highly complementary to CMB, galaxy surveys, and other cosmological probes. In this talk, I will present results from the Carbon monOxide Mapping Array Project (COMAP). COMAP aims to map the large-scale distribution of star-forming matter at redshifts $z = 2$--$3$ and $z = 6$--$8$, respectively constraining the physics of the EoGA and EoR using the CO(1–0) and CO(2–1) lines of CO. In recent years, COMAP has produced two successive world-leading upper limits (Season 1 and 2), and is the first to directly constrain, on the 3D large-scale clustering power spectrum of CO(1-0). The Season 2 constraints between $0.09 \mathrm{Mpc^{-1}} < k < 0.72\mathrm{Mpc^{-1}}$ constrain the CO power spectrum to $kP_\mathrm{CO}(k) < 2400$--$4900 \mathrm{\mu K^2 Mpc^2}$, and is around an order of magnitude improvement on COMAP Season 1. I will present an overview of the COMAP experiment, the state-of-the-art Season 2 analysis and results, as well as preliminary results from the Season 3 analysis.

Speaker: Nils-Ole Stutzer (University of Oslo)

16:00 → 16:30 Coffee

16:30 → 16:45 Forecasting the detection of AME through LiteBIRD

Anomalous Microwave Emission (AME) is a poorly understood Galactic foreground that becomes increasingly relevant for the precision targeted by next-generation CMB experiments, particularly in polarisation where current observations only place sub-percent upper limits on its polarisation fraction. LiteBIRD, with its broad frequency coverage and high sensitivity, offers a unique opportunity to characterise AME in both intensity and polarisation. In this talk I present forecasts of LiteBIRD's ability to detect the AME signal using a template-fitting approach based on cross-correlations between LiteBIRD frequency maps and spatial templates. We find that the recovered amplitude is significantly biased by leakage between spatially correlated foreground components — synchrotron, thermal dust, and AME itself — and I discuss the mitigation strategies under development to recover an unbiased estimate of LiteBIRD's intrinsic sensitivity to AME. I will conclude by outlining how this work will be extended through the inclusion of complementary low-frequency data from ground-based experiments such as QUIJOTE and C-BASS, which is essential to break degeneracies between correlated foregrounds

Speaker: Kumar Aryan (Institute of Astrophysics of Canary Islands)

16:45 → 17:10 Joint Bayesian estimation of the synchrotron sky and modulation kernel

Observations from MeerKLASS exhibit significantly higher chromaticity than standard simulations, requiring approximately much more PCA modes for effective foreground removal than the 4–5 modes typically predicted. To address this discrepancy, we present a Bayesian framework designed to measure and characterize chromatic modulating effects on intensity maps. Our model represents instrumental and environmental effects as a product or convolution of a modulation kernel with the underlying sky signal. By assuming a power-law frequency dependence for the synchrotron sky, while marginalizing over model uncertainties, we simultaneously infer the modulation kernel and its associated statistical uncertainties. We validate this method using simulations that incorporate realistic beam chromaticity and sidelobes. In this talk, I will present preliminary results from this proof-of-concept work, demonstrating the method’s potential for characterize systematics

Speaker: Bruno Bizarria (INPE - Instituto Nacional de Pesquisas Espaciais)

17:10 → 17:15 Summary of the day

Thursday 10 September

09:30 → 10:00 Coffee

10:00 → 10:05 Welcome

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

10:05 → 10:20 Measuring the Coherence Scale of Ultralight Axion-like Particles with Galaxy Polarization

Ultralight axion-like particles (ALPs) generically induce cosmic birefringence through their coupling to photons, but existing measurements based on integrated polarization rotation are primarily sensitive to the overall amplitude of the effect and do not directly probe the spatial coherence of the underlying field. We show that wide-field measurements of galaxy polarization provide a direct probe of the three-dimensional coherence structure of ALPs in the late Universe by exploiting the two-point correlation of the polarization-rotation angle as a function of the physical pair separation $(r)$. If the ALP behaves as nonrelativistic dark matter or a subdominant ultralight component, spatial fluctuations in its oscillating phase generate a correlation signal characterized by a coherence length $(r_{\rm coh}\propto(m_a v)^{-1})$, corresponding to the de~Broglie wavelength of the field. This geometric observable enables a direct and model-independent determination of the ALP mass, effectively breaking the degeneracy between coupling strength and field amplitude inherent in one-point statistics. We construct a tomographic estimator based on three-dimensional pair separations and show that the coherence scale can be robustly extracted from wide-field data using an inverse-variance--weighted collapsed correlation function. For a survey with $(\sim10^6)$ polarized galaxies over a large sky fraction, a polarization rotation amplitude of order $(0.1^\circ)–(1^\circ)$ is detectable at $(5\sigma)$ over the mass range $(m_a\sim10^{-28})–(10^{-27}\,\mathrm{eV})$. The corresponding coherence scale can be measured with a precision $(\sigma(\log_{10} m_a)\lesssim 0.1)$ across a broader range $(m_a\sim10^{-28})–(10^{-26}\,\mathrm{eV})$, provided that the signal is detected. At higher masses $((m_a\sim10^{-26}\,\mathrm{eV}))$, detection becomes challenging for the fiducial survey assumptions, but would be enabled by larger galaxy samples or improved inference of the intrinsic polarization position angles of the sources (e.g., from galaxy morphology). This method provides a direct geometric probe of the coherence scale of ultralight fields, complementary to CMB birefringence and structure-formation constraints.

Speaker: Yasuo Doi (The University of Tokyo)

10:20 → 10:45 Critical Importance of Calibration

High sensitivity is routinely prioritized in observational cosmology, yet the output of precision experiments is ultimately bound by instrument calibration. Too often treated as a late-stage detail, un-optimized calibration strategies—specifically instrumental offsets and gain drifts—directly limit measurement accuracy. Drawing on my experiences with COBE/FIRAS and ARCADE 2, I show how calibration systematics, rather than raw noise, constrained our knowledge of the absolute CMB temperature and bounded the FIRAS spectral distortion limits ($y$). I will show what could have been done differently to address these limitations, tracing how these hard-learned lessons directly shaped the calibration-first design architecture of PIXIE—where absolute calibration is elevated from a secondary routine to the primary mission requirement.

Speaker: Dale Fixsen (University of Maryland/Goddard Space Flight Center)

10:45 → 11:00 Revisiting the CMB spectrum

I will present the current results of the reanalysis of the COBE-FIRAS data, including a new temperature of the CMB monopole and constraints on spectral distortions.

Speaker: Ana Isabel Silva Martins (University of Oslo)

11:00 → 11:30 Coffee

11:30 → 11:55 12 Gyr of CMB–LSS Tomography: From Galactic Dust, tSZ and CIB to CO and [CII]

CMB data are fundamentally projected, with intricate secondary anisotropies each revealing distinct chapters of cosmic evolution. In a series of 6 papers, we carry out a deeply data-intensive cross-correlation tomography of CMB secondaries, jointly analyzing billions of HEALPix pixels across dozens of diffuse sky maps (e.g., Planck, IRAS, Herschel) and spectroscopic galaxies tracing large-scale structure. This unified effort provides a census of the cosmic baryon inventory across 12 Gyr of history. Key results include: (1) measuring the cosmic thermal energy history Ω_th in hot gas through the SZ effect, with implications for gravitational energy release Ω_grav; (2) reconstructing the most precise cosmic star formation history to date via 11-band CIB tomography with fully constrained SEDs over 90% of cosmic time; (3) tracing the evolution of cosmic dust abundance Ω_dust; (4) revealing CIB leakage in Galactic dust maps and developing a new field-level, 3D component separation method, producing the clean “CSFD” dust map; and (5) the first detections of CO and [CII] line backgrounds, probing the molecular gas density Ω_H2 and the cosmic cooling budget. Together, these results point toward a future of end-to-end, multi-wavelength sky analysis in which diffuse backgrounds become precision probes of cosmic evolution.

Speaker: Yi-Kuan Chiang (ASIAA)

11:55 → 12:10 A pixel-based Neural Network method for CMB component separation in polarization.

The detection of the primordial B-mode polarization in the Cosmic Microwave Background (CMB) is one of the most compelling goals in modern cosmology, as it would provide strong evidence for an inflationary period. Achieving this requires accurate component separation techniques to disentangle the cosmological signal from astrophysical foregrounds and instrumental noise. In this work, we present the development of a method based on Neural Networks (NN) to address this challenge. The proposed approach is pixel-based, making it independent of the underlying sky geometry and thus suitable for working directly on spherical data. It is designed to recover the different sky components, including the CMB and polarized galactic components. In the first step, we apply this method to simulated observations from a next-generation CMB experiment, presenting some preliminary results on its performance.

Speaker: Valentina Franco Velásquez (Universidad de Oviedo)

12:10 → 12:35 Component Separation of the Microwave Sky using Information Field Theory

Speaker: Ananya Shankar (Max-Planck Institute for Astrophysics)

12:35 → 13:00 AI/ML Techniques for Model-Independent Cosmological Analysis

Machine learning is increasingly shaping cosmological analyses, though its impact depends critically on careful validation and physically interpretable use. In this talk, I will outline several AI/ML approaches for model-independent cosmological inference, with an emphasis on how data-driven methods can complement more traditional statistical techniques. I will discuss LADDER [1], a deep-learning framework for reconstructing the cosmic distance ladder directly from Type Ia supernovae (SNIa) data while incorporating the full covariance structure of the observations. Following extensive robustness tests, LADDER provides reliable predictions that enable model-independent applications such as consistency checks of baryon acoustic oscillation measurements [2], calibration of high-redshift datasets (e.g., gamma-ray bursts), and the construction of mock catalogues for future SNIa and gravitational-wave (GW) missions. These examples illustrate how carefully validated deep-learning tools can assist cosmological analyses without assuming specific parametric forms. I will also briefly discuss Gaussian-process-based reconstruction of the Hubble parameter and its use in examining the potential cosmological implications of future GW observations [3,4]. Taken together, these methods aim to demonstrate both the potential and the necessary caution, in applying AI/ML techniques to cosmology, and show how responsible non-parametric approaches may offer fresh perspectives on several ongoing challenges in precision cosmology. [1] R. Shah, S. Saha, P. Mukherjee, U. Garain and S. Pal, ApJS 273, 27 (2024) doi:10.3847/15384365/ad5558 [arXiv:2401.17029 [astro-ph.CO]]. [2] R. Shah, P. Mukherjee, S. Saha, U. Garain and S. Pal, [arXiv:2412.14750 [astro-ph.CO]]. [3] R. Shah, A. Bhaumik, P. Mukherjee and S. Pal, JCAP 06, 038 (2023) doi:10.1088/14757516/2023/06/038 [arXiv:2301.12708 [astro-ph.CO]]. [4] P. Mukherjee, R. Shah, A. Bhaumik and S. Pal, Astrophys. J. 960, no.1, 61 (2024) doi:10.3847/15384357/ad055f [arXiv:2303.05169 [astro-ph.CO]].

Speaker: Rahul Shah (Indian Statistical Institute, Kolkata)

13:00 → 14:30 Lunch

14:30 → 16:45 Social afternoon (Details TBA)

Open afternoon for socials before the bowling. We have brought indoor and outdoor games with us, and its also possible to join some other social activities such as walkings tours/scavenger hunts. We will end the afternoon with bowling for everyone.

16:45 → 18:00 Bowling, nachos and drinks

All conference participants are invited to join for bowling and snacks at O'Learys. We have booked 10 lanes including nachos and a drink for everyone. The venue has many other games and food available as well.

Friday 11 September

09:30 → 10:00 Coffee

10:00 → 10:05 Welcome

Conveners: Katrine Alice Glasscock (University of Oslo), Mathew Galloway (University of Oslo)

10:05 → 10:30 Modelling and Forecasting for the FOSSIL and BISOU Experiments

The only precise measurement of the cosmic microwave background (CMB) spectrum to date was achieved by COBE-FIRAS in the early 1990s, demonstrating that the CMB spectrum is extraordinarily close to a perfect blackbody emission. However, both standard and non-standard physical processes are expected to generate small deviations from this spectrum, known as CMB spectral distortions. These distortions provide a unique and largely unexplored probe of the thermal history of the Universe, offering complementary insights into early-Universe physics, particle interactions, and structure formation. The measurement of CMB spectral distortions has therefore become a major objective for future cosmology missions and is now recognised as one of the three key scientific goals of the ESA Voyage 2050 programme. Since 2011, several dedicated space missions have been proposed, FOSSIL being the latest one, having been submitted to the ESA M8 call. In addition, pathfinders projects such as the balloon-borne experiment BISOU (CNES Phase A) are being developed targeting the first CMB spectral distortion monopole detection. After a brief overview of the observational challenges associated with spectral distortion measurements, I will present the latest optimization and forecasting studies for the BISOU and FOSSIL) projects. For this purpose, a flexible spectro-photometric model based on key instrument design choices and subsystem parameters, and a sky emission model dedicated to CMB spectral distortions studies have been developed.

Speaker: Xavier Coulon (IAS-CNRS)

10:30 → 10:55 Spectral distortions beyond the monopole: new targets for CMB imagers

Speaker: Jens Chluba (University of Manchester)

10:55 → 11:10 Beyond the BLUE: CRLB-Referenced Benchmarking of Neural vs. Linear Estimation in mm/submm Surveys

Convolutional Neural Networks (CNNs) are now widely deployed in astronomy, often claiming superior performance over classical linear estimators for an extensive range of data applications. Yet it remains unclear when a CNN can genuinely outperform the matched filter (MF) -- which attains the Cramér–Rao Lower Bound (CRLB) as the minimum-variance unbiased estimator (MVUE) under Gaussian noise, and remains the best linear unbiased estimator (BLUE) in general. We test a ResNet-based CNN against the MF using image noise properties as the sole performance discriminator, and ask: when does a CNN's advantage reflect genuine information beyond the MF, rather than the bias-variance tradeoff inherent in any trained regression model? We address this through a series of realistic noise scenarios encountered in mm/submm astronomy (e.g., CMB surveys), demonstrating the surprising robustness of matched filtering, and isolating two distinct regimes where a CNN offers genuine advantage. The first is the conditionally Gaussian regime: per-image covariance variation that opens a modest CNN edge through adaptive noise modeling. The second is the genuinely non-Gaussian case, where the CNN exploits higher-order noise statistics to achieve lower-variance estimation than any linear filter. We further show that the advantage is template-dependent: governed by the spectral overlap between the source profile and the noise departure from Gaussianity, with distinct scaling for compact versus extended sources.

Speaker: Kaustuv Basu (Bonn University)

11:10 → 11:40 Coffee

11:40 → 12:05 Achievements and the Current Status of MAXI

MAXI (the All-Sky X-ray Imager) is an all-sky X-ray monitor that has been in operation on the International Space Station (ISS) since 2009. I will provide an overview of MAXI’s achievements over the past 17 years, including the discovery of 17 X-ray novae, and its current status.

Speaker: Ken Ebisawa (ISAS/JAXA)

12:05 → 12:20 In Search of Gravitational Waves from Fast X-ray Transients

Fast X-ray Transients (FXTs) are brief, intense X-ray flashes unassociated with persistent X-ray sources or known stellar objects, typically lasting from seconds to hours. Previous studies have identified the origins of some FXTs, e.g., XRO 080109/SN2008D as supernova shock breakouts and EP240315a as longduration Gamma-Ray Bursts. In the past 4 decades, a portion of FXTs has been identified as Gamma-Ray Bursts (GRBs). However, the origins of the rest of the FXTs remain uncertain, prompting the development of theoretical models, including binary neutron star mergers, accretion-induced collapse of compact objects, and tidal disruption events. With the launch of Einstein Probe in January 2024, the study of FXTs has entered a new phase. Its large field of view significantly increases the discovery rate, enabling a more robust statistical analysis. As of April 2025, Einstein Probe has reported approximately 80 FXTs — about four times the 22 events identified by Chandra between 2000 and 2022. We propose a methodology to search for possible relations between Gravitational Waves and FXTs. To achieve this, we intend to utilize X-Pipeline, an unmodeled search algorithm to identify excess coherent signals in LIGO-Virgo-KAGRA (LVK) O4 observations. By applying minimal assumptions, the search aims to test the existence of any temporal and spatial correlation, which may provide new insights into the physical mechanisms driving these transient X-ray events.

Speaker: Po-Ya Wang (National Tsing Hua University)

12:20 → 12:45 The state of Commander4

The Cosmoglobe collaboration has over the last decade developed and utilized the Commander3 Fortran codebase for global Bayesian CMB analysis. Commander4 is a ground-up rewrite of this algorithm with the necessary performance and scalability to analyze modern (>10TB) CMB datasets. While less mature than Commander3, it has now matured to the state of being able to perform preliminary analysis of datasets like SO and LiteBIRD simulations.

Speaker: Jonas Lunde (University of Oslo)

12:45 → 13:00 Next steps: Cosmoglobe Xtreme

Speaker: Mathew Galloway (University of Oslo)

13:00 → 13:15 Thank you -- get involved!

13:15 → 14:45 Lunch