This is the COSI internal release. The public release is planned for April 1, 2025. Some things are still missing or incomplete, as specified below:
- The SAA background component is not available. This is highly computationally intensive (requiring ~2,739 core years), and we are currently working on completing this using both the Mainz and NASA clusters. Correspondingly, the total background file is not available yet, and the page describing the background simulations for DC3 is incomplete.
- Response files need to be converted from standard .h5 files to .h5 files with good chunks. The h5 files with good chuncks will optimize the speed and memory usage.
- No binned data products have been provided. Binning the data is part of the standard analysis procedure. However, in some cases, binning the data may require lots of RAM, making it difficult for standard laptops (as was found for DC2). Please let us know if you are running into related difficulties, and we can see if we need to provide some binned data products.
- A small number of the source simulations have not been completed. Generally, these are input models that require a bit of extra work to implement in the simulation pipeline. We will be working to complete the remaining models before the public release. If you notice that your model is missing and were planning to test it in the near future, please let us know and we will make it a priority.
- The polarization analysis tools do not yet account for the instrument orientation. Consequently, only GRB polarization on short time scales can be analyzed at the moment. This is a top priority that we are working on updating.
- The source injector only works for point sources, and it doesn't account for Earth occultation. The Earth occultation will be implemented soon, and we are also updating the code to allow for injecting extended sources. Both of these should be ready by the public release.
- The description of each data challenge needs to be checked for content by the science team member that provided it. You will find notes at the top of each data challenge with the requested information.
- Introduction
- Getting Started
- System Requirements
- Getting Help
- Backgrounds
- Earth Occultation
- Polarization
- Releases
- Computing Resources
- Simulation Tools
- Summary of Current and Past Challenges
- Useful Reference Guides
- Data Challenges
- Known Caveats and Limitations
- Citing
Welcome to the third COSI Data Challenge (DC3)! The COSI Data Challenges are released on a yearly basis in preparation for the launch of the COSI Small Explorer (SMEX) class mission in 2027 (Tomsick+23). They are based on simulated data, which is intended to closely mimic the real flight data. Every year the Data Challenges have increasingly more realistic source and background models, and they are analyzed with increasingly complete and matured analysis tools. In general there are two main goals of the Data Challenges:
- Facilitate development of the COSI data pipeline and analysis tools
- With routine feedback from scientists
- Alongside development of the expected source models by the science team
- Provide resources to the astrophysics community to become familiar with COSI data
- Excellent training for science team in preparation for first analyses after launch
- Public releases help with community building before COSI data is released
The only software requirement for DC3 is cosipy. A general introduction into cosipy, including installation instructions, can be found in the cosipy-intro directory. For a general introduction into analyzing data from Compton telescopes see Compton-telescope-data-analysis-intro. Note that cosipy is part of the larger COSITools, which is a broad collection of COSI data analysis tools, documentation, and verification data sets. COSITools can be installed by following the installation instructions here. This also includes MEGAlib, which is the main software program used for running simulations. However, unless you need MEGAlib and/or COSITools for other reasons, it's highly recommended to just install cosipy.
This year's Data Challenge is based on 3 months of exposure time, for an equatorial orbit at an altitude of 530 km, with a pointing that rocks between
The input models and challenges for DC3 were provided by the COSI science teams. There are challenges for the different science groups: GRBs, Positrons, Nucleosynthesis, Galactic, Extragalactic, and Dark Matter. These are described in detail in the Data Challenges section below.
Users are encouraged to post feedback on the Discussions page. This can include solutions to specific challenges, questions, issues, etc.
In summary, to get started with DC3, install cosipy, familiarize yourself with the data-products, and then start working through the Data Challenges, as described below.
One of our goals in developing cosipy is to make it easily accesible to all users. All of the Data Challenges starting with DC2 should be doable on a laptop with at least 16 GB of RAM. We are still working on optimizing the code, and so please let us know if you are running into memory issues.
Please submit a new issue in the cosipy git repository if you have issues with the code. If you have general feedback, or need further assistance, please reach out to the COSI Data Challenge team lead, Chris Karwin ([email protected]), the cosipy implementation lead, Israel Martinez-Castellanos ([email protected]), and the pipeline development lead Carolyn Kierans ([email protected]).
In general, observations in the MeV band are hindered by high backgrounds (both instrumental and astrophysical). In order to ensure that COSI accomplishes its main science goals, it is therefore crucial to have a firm understanding of these backgrounds. DC3 includes all of the background components. Compared to the background estimates from DC2, we have now included the full SAA passage, as well as the Galacic diffuse continuum emission. Further details can be found in the backgrounds directory.
For analyzing data in DC3, the starting point is to model the backgrounds using the actual simulated backgrounds themselves. This is the ideal case, where the backgrounds are perfectly known, which of course is not very realistic. The next step is to estimate the backgrounds. With DC3 we have new methods to estimate the background for both line and continuum sources. The background estimation tool for line emission can be used for both point sources and extended sources. An example tutorial can be found in the line background estimation notebook. The background estimation tool for continuum emission is currently only available for point sources. An example tutorial is available in the continuum background estimation notebook. We stress that both of these background estimation algorithms are only first versions, and further development and testing is still needed. More details are provided in the respective example tutorials.
The available background files are listed in the data-products directory. We provide a file with the total background, as well as separate files for the individual background components (as details in the backgrounds directory).
The Earth blocks a significant portion of the sky for satellites in low-Earth orbit, referred to as Earth occultation. It is important to account for this when simulating observations and performing data analysis. In order to implement this for the DC3 simulations we added new functionality to MEGAlib (develop-cosi branch), as detailed in the earth-occultation directory. These new methods now allow for simulating instruments with non-zenith pointings. We also added new methods in cosipy to account for Earth occultation in the data analysis.
Polarimetry is a key aspect of COSI's primary science goals. With DC3 we release the first version of our polarization tools in cosipy. We also added new functionality in MEGAlib to define polarization in Galactic coordinates. This allows for simulating polarized sources together with the instrument's orbit. A general introduction into Compton polarimetry can be found in the polarization directory.
- Data challenge 1, March 2023: cosi-data-challenge-1
- Data challenge 2, March 2024: cosi-data-challenge-2
- Data challenge 3, April 2025: cosi-data-challenge-3 (latest release)
- Data challenge 4: Planned for March 2026
- Data challenge 5: Planned for March 2027 (final challenge before launch 🚀!)
The simulations for the COSI Data Challenges are run on high performance computing clusters. Most notably, we have made extensive use of NASA's Discover cluster, the MOGON cluster in Mainz, and Clemson University's Palmetto cluster.
The simulations employ MEGAlib via the Python-based COSI simulation pipepline, cosi-sim. Details regarding the specific MEGAlib versions and configuration files can be found in each respective Data Challenge directory. Model inputs for the simulations and the corresponding Data Challenges come from the COSI science team. All of the models used for past Data Challenges can be found in the source library of the cosi-sim tools (link).
- Data Challenge 1:
- Focused on the 2016 COSI Balloon flight.
- Release includes real flight data for the Crab.
- Main goal is to learn the fundamentals of analyzing Compton data with COSI.
- The analysis tools used for DC1 are only preliminary (referred to as cosipy classic).
- Developed by Thomas Siegert for analysis of the 2016 balloon data.
- Contains 3 straightforward examples of COSI’s science goals:
- Extracting energy spectra from the Crab, Cen A, Cygnus X-1, and Vela.
- Imaging bright point sources, such as the Crab and Cygnus X-1.
- Imaging diffuse emission from the positron-electron annihilation 511 keV and Al-26 1.8 MeV gamma-ray lines.
- Data Challenge 2:
- Focused on COSI SMEX mission.
- First (alpha) release of cosipy.
- Data challenges for all the main science groups (none for dark matter and solar).
- All models and challenges provided by respective COSI science teams.
- Uses 3 months of observations, for an equatorial orbit at 550 km, with a zenith pointing.
- All BG components are included, except for SAA passage (i.e. trapped cosmic rays).
- BG also includes time variability from changing geomagnetic cutoff.
- We simulated 12 background components, and 30 unique sources, running 49 different source simulations in total (using multiple models for some of the sources).
- Contains 7 main tutorials demonstrating all the tools/methods needed for completing the challenges, included as part of the cosipy release:
- dataIO
- GRB localization
- GRB spectral fit
- Crab spectral fit
- 511 spectral fit
- Crab imaging
- 511 imaging
- Data Challenge 3
- Focused on COSI SMEX mission.
- Data challenges for all the main science groups (including dark matter), covering all of COSI's primary science objectives.
- All models and challenges provided by respective COSI science teams.
- Used 3 months of observations, for an equatorial orbit at 530 km.
- Simulations include rocking of instrument:
- Pointing changes between +/- 22 degrees every 12 hrs, with 8 minute transition time.
- Used detailed COSI SMEX mass model.
- Simulated all background components in low-Earth orbit, including variability from geomagnetic cutoff, long-term buildup, and full SAA passage.
- Background includes the Galactic diffuse continuum for the first time.
- New methods in both MEGAlib and cosipy to account for Earth occultation with a non-zenith pointing.
- First time including polarization.
- Numerous improvements to cosipy:
- First version of source injector.
- New implementation of Earth occultation in detector response.
- First polarization tools.
- New methods to estimate the background for continuum sources and line sources.
- Refinements and further developments of imaging class.
- New Extended source response class.
- Introduction to Compton telescope data analysis
- Introduction to cosipy
- Summary of background simulations
- Dealing with Earth Occultation
- Introduction to Polarization
We have created example Jupyter notebooks demonstrating all of the tools that will be needed to complete this year's data challenges. They are available as part of the cosipy release, and listed below:
Example 1: dataIO
Example 2: GRB localization
Example 3: GRB spectral fit
Example 4: Crab spectral fit
Example 5: 511 spectral fit
Example 6: Crab imaging
Example 7: 511 imaging
Example 8: Source injector
Example 9: TS maps
Example 10: Polarization (ASAD method)
Example 11: Continuum background estimation
Example 12: Line background estimation
If you haven't worked with Jupyter before, you can find some help here.
As a very first step, try working through some of the example notebooks. Specific challenges for the different science topics are described below. You can start with whichever topic you are most interested in. Each challenge will refer you to a specific example notebook that will demonstrate the basic tools needed to complete the respective challenge. If you have completed the main challenges and are interested in exploring other models, you can employ the source injector (see the Source injector example). If you are interested in getting more involved in the cosipy development, see the Known Caveats and Limitations section at the bottom of this page, which outlines some of the priority areas for the next stages of development.
All input models used for the simulations can be found in the DC3 source library of the COSI simulation pipeline, available here. This includes all the information about the injected sources, and it can be used for checking the results of the data challenges.
Orientation Files:
Two orientation files are available:
DC3_final_530km_3_month_with_slew_15sbins_GalacticEarth_SAA.ori
DC3_final_530km_3_month_with_slew_1sbins_GalacticEarth_SAA.ori
The 1 second binning may be optimal for analyzing transients on short time scales, but generally the 15 second binning should be sufficient and is considered the default.
Background Files:
There are a few different options for modeling the background. The staring point is to use the ideal case, where the background model in the analysis is the same as the simulated background. We have provided a file with the total background, as well as files for the individual background components. To simplify the analysis, it is sometimes helpful to start with just a single background component (e.g. albedo photons), and then move on to the total background after everything is working. A more realistic estimate of the uncertainty on the background modeling can be achieved by using one of the background estimation tools (see the Continuum background estimation and Line background estimation examples). However, we caution that these tools are in thier very early stages, and they still require further testing and development.
The tools needed to complete these challenges are demonstrated in the GRB spectral fit, GRB localization, and Polarization (ASAD method) examples.
Data Files:
ResponseContinuum.o3.e100_10000.b10log.s10396905069491.m2284.filtered.nonsparse.binnedimaging.imagingresponse_nside8.area.h5.gz
ResponseContinuum.o3.pol.e200_10000.b4.p12.s10396905069491.m441.filtered.nonsparse.binnedpolarization.11D_nside8.area.h5.gz (for polarization)
GRB_bn081207680_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn090424592_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn100612726_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn110605183_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn131122490_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn140329295_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn161004964_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn170405777_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn180504136_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_bn180703876_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF051103_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF070201_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF070222_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF180128A_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF200415A_3months_unbinned_data_filtered_with_SAAcut.fits.gz
GRB_MGF231115A_3months_unbinned_data_filtered_with_SAAcut.fits.gz
MgtBurst_bright_complex_3months_unbinned_data_filtered_with_SAAcut.fits.gz
MgtBurst_bright_simple_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
All input models can be found here.
The GRBs occur randomly within the orientation file, with their positions chosen such that they have incidence angles under
- bn081207680: PF = 0, PA =
$0^\circ$ , t = 1836496300.0 s - bn090424592: PF = 0.1, PA =
$110^\circ$ , t = 1837507002.0 s - bn100612726: PF = 0.2, PA =
$35^\circ$ , t = 1839617230.0 s - bn110605183: PF = 0.3, PA =
$50^\circ$ , t = 1841896404.0 s - bn131122490: PF = 0.4, PA =
$175^\circ$ , t = 1842123855.0 s - bn140329295: PF = 0.5, PA =
$95^\circ$ , t = 1837652924.0 s - bn161004964: PF = 0.6, PA =
$10^\circ$ , t = 1842948485.0 s - bn170405777: PF = 0.7, PA =
$160^\circ$ , t = 1841915858.0 s - bn180504136: PF = 0.8, PA =
$45^\circ$ , t = 1836985181.0 s - bn180703876: PF = 0.9, PA =
$25^\circ$ , t = 1838652949.0 s
Information for the MGFs, including reference papers (PF = 1, PA =
- MGF051103 (Svinkin+21): t = 1835533723.498 s
- MGF070201 (Ofek+08): t = 1835537496.0 s
- MGF070222 (Svinkin+16): t = 1835491681.0 s
- MGF180128A (Trigg+24): t = 1835488380.0 s
- MGF200415A (Svinkin+21): t = 1835488530.0 s
- MGF231115A (Trigg+25): t = 1835533696.0 s
Information for magnetar short burst:
- MgtBurst_bright_complex: no polarization, t = 1835640345.022513 s
- SGR 1935+2154 (Li+21) - a bright burst with a complex light curve.
- MgtBurst_bright_simple: no polarization, t = 1837365120.031250 s
- A bright burst with simple light curve: spectrum of SGR 1935+2154 burst + light curve of 1E 1048.1-5937 (Gavriil+02) burst.
GRB and MGF Goals:
- Localize GRB
- Fit spectrum
- Measure polarization (fraction and angle)
- Classify: GRB or MGF
Magnetar Short Burst Goals:
- Check if they are detectable
- If detectable, localize and fit spectrum (and polarization when applicable)
The tools needed to complete these challenges are demonstrated in the 511 imaging and 511 spectral fit notebooks.
All challenges should use the same detector response files:
Response511.o4.e509_513.s20881894470591.m2555.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
ResponseContinuum.o3.e100_10000.b10log.s10396905069491.m2284.filtered.nonsparse.binnedimaging.imagingresponse_nside8.area.h5.gz
extended_source_response_511_merged.h5.gz (precomputed 511 extended source response file)
extended_source_response_continuum_merged.h5.gz (precomputed continuum extended source response file)
The line response is for analyzing the 511 keV line emission, and the continuum response is for analyzing the orthopositronium continuum. Currently, these two components cannot be analyzed simultaneously, as desribed in the Known Caveats and Limitations section.
- Proofread/check content
- Provide links to cited papers
Data Files:
LMC_Gaussian_511_3months_unbinned_data_filtered_with_SAAcut.fits.gz
LMC_Gaussian_511_x100_3months_unbinned_data_filtered_with_SAAcut.fits.gz
M31_Gaussian_511_3months_unbinned_data_filtered_with_SAAcut.fits.gz
M31_Gaussian_511_x100_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Virgo_Gaussian_511_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Virgo_Gaussian_511_x100_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
Fluxes are estimated by assuming that the 511 keV photon flux is proportional to the stellar mass of a source. Using a Milky Way 511 keV flux of
The stellar masses are:
LMC:
M31:
Virgo:
We also include each source at 100x the nominal flux, in order to ensure that they are above COSI's 511 keV line sensitivity.
Goals:
- Determine COSI's sensitivity for detecting these potential extragalactic 511 keV sources.
- Clarify and provide a bit more information regarding how the 511 flux was estimated.
- Proofread/check content
- Provide links to cited papers
Data Files:
Globular_Cluster_Tuc_47_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Globular_Cluster_Omega_Cen_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Globular_Cluster_NGC_6397_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Globular_Cluster_NGC_6121_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The 511 keV photon flux for the brightest 4 globular clusters is presented here. The flux is estimated based on the 0.1 - 100 GeV luminosity from the Globular clusters, nuclear bulge emission, and boxy bulge emission in the Milky Way (Bartels+2017). The GeV luminosity of the Bulge is compared to the 511 keV luminosity of the bulge to estimate the 511 keV flux of the globular clusters. The flux values shown here are 3x the values estimated from the correlation in order to make them within COSI's sensitivity limit.
Goals:
- Determine COSI's sensitivity for detecting 511 keV emission from Globular clusters.
- Proofread/check content
Data Files:
Positrons_from_26Al_line_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Positrons_from_26Al_cont_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
This is the description of the model of the diffuse emission due to the steady state annihilation of positrons produced by 26Al decay in our Galaxy. The spatial distribution is derived from the calculation of the propagation of positrons in our Galaxy. The initial positions of positrons follow the model NE2001 of Cordes & Lazio 2002 that fits the distribution of massive stars in our Galaxy, with a fraction of 26Al in the bulge of 2%. The positron propagation method is described in Alexis et al. 2014. The spectral distribution takes into account the annihilation line and the orthopositronium continuum. It was obtained from the model of Guessoum et al 2005 that compute a spectrum for each phase of the interstellar medium. The spectral model was corrected by the Galactic rotation using the model of Fich, Blitz and Stark 1989 for R > 3 kpc and a solid rotation model for R < 3 kpc.
Goals:
- Detection of the diffuse line and continuum emissions
- Detection of the Doppler shift in the disk
- Detection of the spectral shape
- Correlation with the 26Al map (1809 keV emission)
- Proofread/check content
Data Files:
Positrons_from_44Ti_line_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Positrons_from_44Ti_cont_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
This is the description of the model of the diffuse emission due to the steady state annihilation of positrons produced by 44Ti decay in our Galaxy. The spatial distribution is derived from the calculation of the propagation of positrons in our Galaxy. The initial positions of positrons follow the model NE2001 of Cordes & Lazio 2002 that fits the distribution of massive stars in our Galaxy, with a fraction of 44Ti in the bulge of 2%, which corresponds to the fraction of massive stars also used for the 26Al. The positron propagation method is described in Alexis et al. 2014. The positron rate due to 44Ti decay is
Goals:
- Detection of the diffuse line and continuum emissions
- Detection of the Doppler shift in the disk
- Detection of the spectral shape
The tools needed to complete these challenges are demonstrated in the 511 imaging and 511 spectral fit notebooks.
- Proofread/check content
Data Files:
Response26Al.o4.e1805_1812.s10036231691364.m1045.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
extended_source_response_Al26_merged.h5.gz (precomputed extended source response file)
26Al_Cyg_Region_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The characteristics of the emission were obtained from the analyses of INTEGRAL/SPI observations
described in Martin+09. The source is modeled with a
Goals:
- Make detection taking into account the Galactic diffuse continuum background at 1809 keV emission
- Measure width of the gamma-ray line
- Recover 60Fe/26Al ratio (see 60Fe_Cyg_Region)
- Proofread/check content
Data Files:
Response26Al.o4.e1805_1812.s10036231691364.m1045.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
extended_source_response_Al26_merged.h5.gz (precomputed extended source response file)
26Al_NE2001_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
This is the description of the model of the 1809 keV line diffuse emission due decay of 26Al in our Galaxy. The spatial distribution is derived from the model NE2001 of Cordes & Lazio 2002 that fits
the distribution of massive stars in our Galaxy. The fraction of 26Al in the bulge is 2%. This value is a compromise between SPI observations and results of studies of star formation rate in this region.
The shape of the line in each spatial bin takes into account the turbulence of the interstellar medium and the Galactic rotation using the model of Fich, Blitz and Stark 1989 for R > 3 kpc and a solid rotation model for R < 3 kpc.
Goals:
- Detection of the diffuse emission
- Detection of the Doppler shift in the disk
- Detection of the spectral shape
- Correlation with the emission of the Galactic positron annihilations
- Extract F(26Al)/F(60Fe) ratio and its uncertainty
The tools needed to complete these challenges are demonstrated in the Crab spectral fit and GRB localization notebooks.
- Clarify and provide a bit more information regarding the models
- State the goals of the data challenge
- Proofread/check content
- Provide links to cited papers
Data Files:
Response44Ti.o4.e1154_1160.s9607532021290.m1215.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
extended_source_response_Ti44_merged.h5.gz (precomputed extended source response)
CasApartiallyresolved_3months_unbinned_data_filtered_with_SAAcut.fits.gz
CasAfullyresolved_3months_unbinned_data_filtered_with_SAAcut.fits.gz
CasAG16distribution_3months_unbinned_data_filtered_with_SAAcut.fits.gz
CasAunresolved_3months_unbinned_data_filtered_with_SAAcut.fits.gz
CasAsymmetric_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
In all 4 scenarios, bulk center of motion is at rest. Doppler broadening is limited to 1000 km/s, as otherwise, when combined with the Doppler shifts of clumps, the signal will fall quite outside the DC2 simulated response region of 1143-1171 keV.
In asymmetric scenarios, Clump 1: Contains 2/3 of total 44Ti yield Doppler shifted towards from the observer (blueshifted). Has peak energy higher than 1157 keV. Clump 2: Contains 1/3 of total 44Ti yield Doppler shifted away from the observer (redshifted). Has peak energy lower than 1157 keV.
All spectra follow simple Gaussian distributions. The flux is taken as the value between
Goals:
- What are the goals?
- Proofread/check content
Data Files:
Response60FeHigh.o4.e1329_1336.s10201526728102.m1287.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
Response60FeLow.o4.e1170_1176.s9552269354945.m1188.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
extended_source_response_Fe60_low_merged.h5.gz (precomputed extended source response)
extended_source_response_Fe60_high_merged.h5.gz (precomputed extended source response)
60Fe_Cyg_Region_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The flux of the 60Fe lines is derived from the calculation of Martin+10. The spatial extent of this diffuse emission is the same as the one of the 1809 keV line emission that were obtained from the analyses of SPI/INTEGRAL observations
described in Martin+09. The source is modeled with a 3deg width (standard deviation) Gaussian shape emission centered at
Goals:
- Make detection taking into account the Galactic diffuse continuum background at 1173 keV and 1332 keV
- Measure width of the gamma-ray line
- Recover 60Fe/26Al ratio (see 26Al_Cyg_Region)
- Proofread/check content
Data Files:
Response60FeHigh.o4.e1329_1336.s10201526728102.m1287.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
Response60FeLow.o4.e1170_1176.s9552269354945.m1188.filtered.nonsparse.binnedimaging.imagingresponse_nside16.area.h5.gz
extended_source_response_Fe60_low_merged.h5.gz (precomputed extended source response)
extended_source_response_Fe60_high_merged.h5.gz (precomputed extended source response)
60Fe_NE2001_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
This is the description of the model of the diffuse emission of the 1173 keV and 1332 keV lines due decay of 60Fe in our Galaxy. The spatial distribution is derived from the model NE2001 of Cordes & Lazio 2002 that fits the distribution of massive stars in our Galaxy. The fraction of 60Fe in the bulge is 2%. This value is the same as the one of the 26Al, which is a compromise between SPI observations and results of studies of star formation rate in this region. The flux is computed assuming a total mass of 60Fe of 3.5 M_sol in our Galaxy (see Wang+20 and Siegert+23). The shape of the line in each spatial bin takes into account the turbulence of the interstellar medium and the Galactic rotation using the model of Fich, Blitz and Stark 1989 for R > 3 kpc and a solid rotation model for R < 3 kpc.
Goals:
- Detection of the diffuse emission.
- Extraction of the F(26Al)/F(60Fe) ratio and its uncertainty.
The tools needed to complete the Galactic challenges are demonstrated in the Crab spectral fit, Crab imaging, GRB localization, Polarization (ASAD method), and 511 spectral fit (for extended source analysis) notebooks.
All challenges should use the same detector response file:
ResponseContinuum.o3.e100_10000.b10log.s10396905069491.m2284.filtered.nonsparse.binnedimaging.imagingresponse_nside8.area.h5.gz
ResponseContinuum.o3.pol.e200_10000.b4.p12.s10396905069491.m441.filtered.nonsparse.binnedpolarization.11D_nside8.area.h5.gz (polarized sources)
extended_source_response_continuum_merged.h5.gz (precomputed extended source response file)
Data Files:
GalTotal_SA100_F98_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The Galactic diffuse continuum emission is modeled using the v57 release of the GALPROP cosmic ray (CR) propagation and interstellar emissions framework (Porter+22). GALPROP self-consistently calculates spectra and abundances of Galactic CR species and associated diffuse emissions (gamma rays, X-rays, and radio) in 2D and 3D. The v57 release includes a set of steady-state emission model examples that reproduce the latest CR data. There are six models in total, categorized according to the CR source and interstellar radioation field (ISRF) model used for the prediction. There are 3 CR source models (SA0, SA50, SA100) and two ISRF models (R12, F98). The CR source density models are based on the distribution of injected CR power, with SA0 describing an axisymmetric disk (following the radial distribution of pulsars), SA50 describing a 50/50% split of the injected CR luminosity between disk-like and spiral arms, and SA100 describing pure spiral arms. All models have the same exponential scale height of 200 pc. The two ISRF models employ different spatial densities for both the stars and the dust but produce intensities very similar to those of the data for near- to far-infrared wavelengths at the location of the solar system (see Porter+17 and references therein). For the neutral gas distributions (atomic and molecular), a 3D model from Johannesson+18 is employed. These GALPROP models include the total emission, which is dominated by inverse Compton radiation, but also has a small contribution from Bremsstrahlung towards the upper energy bound. As our representative case, for DC3 we simulate the SA100-F98 model.
This is the first data challenge to include the Galactic diffuse continuum, and our corresponding goals this year are straight forward. Future data challenges will look more into the different model variations and key parameters.
Note that the Galactic diffuse continuum emission is also part of the standard background model for COSI, which will be employed for most analyses.
Goals:
- Measure the spectrum of the Galactic diffuse continuum emission, extracting it from the rest of the background
- Image the Galactic diffuse continuum emission in the COSI energy band
- Characterize how the the Galactic diffuse continuum emission impacts the sensitivity for point sources in the Galactic plane
- Proofread/check content
Data Files:
PSRB1259_3months_unbinned_data_filtered_with_SAAcut.fits.gz
PSRB1259_10x_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
PSR B1259-63 is a binary gamma-ray system consisting of a radio pulsar and a massive Be-type star. The system has a highly eccentric orbit, with an orbital period of 3.4 years. Gamma-ray emission occurs during its periastron passage, as interactions between the outflows of the two objects trigger particle acceleration. The next periastron passage will be November 19th, 2027, making this a prime target for COSI. The model for DC3 is based on the work in Abdo+11 (see model 1 in the bottom of Figure 5). For the DC3 simulations, the following features are assumed: 1) Constant emission for 30 days. 2) Two flux scenarios: the nominal value from Abdo+11, as well as a 10x enhanced flux. Note that in general the flux and duration vary for each periastron passage, and in COSI's energy band they are not well understood.
Goals:
- Measure MeV gamma-ray flux during the flare event.
- Determine the duration of flare periods.
- Proofread/check content
Data Files:
GRS175_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
Best fit comptonization model of epoch2 of Pottschmidt+06.
Goals:
- A simple test of spectral measurement for this galactic center source.
- Proofread/check content
Data Files:
cygX1_soft_3months_unbinned_data_filtered_with_SAAcut.fits.gz
cygX1_hard_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The two spectral models for Cyg X1 are best fit eqpair models of time averaged INTEGRAL data Cangemi+21 given for hard and soft states respectively. Polarization is also included. The hard state model is based on the measurements of Rodriguez+2015. At low energy (0.1 - 0.4 MeV) the polarization fraction is 5% with an angle of 40 degrees (IAU convention). At high energy (0.4 - 10 MeV) the polarization fraction is 75% with the same angle. The soft spectral state assumes an energy-independent polarization of 20% (same angle).
Goals:
- Check detection sensitivty in soft state
- Test how well COSI can monitor for spectral transitions
- Proofread/check content
Data Files:
1E1740_compow_3months_unbinned_data_filtered_with_SAAcut.fits.gz
1E1740_twocompt_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The 2 spectral models for 1E1740.7-2942 (also know as great annihilator) are the best fit models
of INTEGRAL data obtained by Bouchet+09:
- compow: thermal comptonization + power law
- twocompt: two components of thermal comptonization with different temperatures Both models represent the INTEGRAL data well but strongly differ at the highest energies.
Goals:
- Test whether COSI would be able to distinguish between the two models.
The tools needed to complete the Extragalactic challenges are demonstrated in the Crab spectral fit and Polarization (ASAD method) notebooks.
All challenges should use the same detector response file:
ResponseContinuum.o3.e100_10000.b10log.s10396905069491.m2284.filtered.nonsparse.binnedimaging.imagingresponse_nside8.area.h5.gz
ResponseContinuum.o3.pol.e200_10000.b4.p12.s10396905069491.m441.filtered.nonsparse.binnedpolarization.11D_nside8.area.h5.gz (polarized sources)
extended_source_response_continuum_merged.h5.gz (precomputed extended source response)
- Proofread/check content
- Provide links to cited papers
Data Files:
NGC_1068_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The baseline model is a powerlaw with exponential cut-off from Bauer+2015:
Gamma=1.92, Ecut=200 keV; intrinsic flux 2-10 keV = 8.9e-10 erg/cm2/s
Goals:
- Determine flux in the COSI band, and coronal cut-off energy
- Proofread/check content
- Provide links to cited papers
Data Files:
NGC_4151_bright_3months_unbinned_data_filtered_with_SAAcut.fits.gz
NGC_4151_EC200_3months_unbinned_data_filtered_with_SAAcut.fits.gz
NGC_4151_EC1000_3months_unbinned_data_filtered_with_SAAcut.fits.gz
NGC_4151_faint_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The baseline model is a powerlaw with exponential cut-off.
flux in the 20-30 keV calibrated from NuSTAR observations:
NGC_4151_ec200: Gamma=1.75, Ecut=200 keV
NGC_4151_ec1000: Gamma=1.75, Ecut=1000 keV
flux calibrated from INTEGRAL observation of Lubinski+2010:
NGC_4151_bright: Gamma=1.71, Ecut=264 keV
NGC_4151_faint: Gamma=1.81, Ecut=1000 keV
Goals:
- Determine flux in the COSI band, and coronal cut-off energy
- Give gaol of challenge
- Provide description of model
- Proofread/check content
Data Files:
4C21p35_noflare_3months_unbinned_data_filtered_with_SAAcut.fits.gz
4C21p35_flare_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
We present a lightcurve showing 2 states: a flaring state and a quiescent state. The Flaring state is defined every time in which the average flux is 3 times greater than the 16-years average flux (given by Fermi).
The two states come with two different spectra: both powerlaws with different indices:
- noflare = 1.6
- flare = 2.5
The normalization is derived from the integrated flux in COSI energy band derived from the extrapolation of the Fermi-LAT log parabola function.
Goals:
- What are the goals of this challenge?
Data Files:
3C279_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
The spectral data is for 3C 279 high, which represent the high state of the source, and the flux is increased by 100x its nominal value. The source is polarized with a polarization fraction of 19.62%, and a randomly chosen polarization angle of 45 degrees (in IAU convention).
Goals:
- Measure the polarization fraction and angle.
- Provide more info about models.
- Please let us know what kind of spectra you have (line and/or continuum). This will determine the type of respsonse needed.
- Proofread/check content
The tools needed to complete these challenges are demonstrated in the 511 imaging and 511 spectral fit notebooks.
Data Files:
eeg_ISO_3months_unbinned_data_filtered_with_SAAcut.fits.gz
eeg_Bur_3months_unbinned_data_filtered_with_SAAcut.fits.gz
eeg_NFW_3months_unbinned_data_filtered_with_SAAcut.fits.gz
gg_ISO_3months_unbinned_data_filtered_with_SAAcut.fits.gz
gg_Bur_3months_unbinned_data_filtered_with_SAAcut.fits.gz
gg_NFW_3months_unbinned_data_filtered_with_SAAcut.fits.gz
Input Models:
Files for DM annihilating into gamma-gamma or e-e-gamma, assuming NFW or Burkert profile.
m_DM = 3 MeV and = 1e-30 cm3/s. Other parameters are detailed in our slides.
Goals:
- Calculate the gamma-ray spectra from the annihilating MeV DM.
- Compare the spectra from DM and background, and determine COSI’s detectability for attractive DM candidates.
The items listed here are some of the priorities for DC4 development. These can be considered as extra/advanced challenges, and anybody is welcomed to work on them, with the ultimate goal of implementing the software solutions into cosipy.
- It is not currently possible to simultaneously fit continuum and line components. We have separate response files for different emission components (i.e. continuum, 511 keV, Aluminum-26, etc.), and with the current binned analysis setup in cosipy, the data binning needs to match the response binning, and thus only a single component can be analyzed at a time. Possible solutions to this include:
- Creating a single response for all components
- Creating a class that automatically matches an input model with the corresponding response
- Reparameterizing the response in such a way that prevents this issue
- The background estimation tools need to be further tested and developed. With DC3 we have provided first versions for estimating continuum and line backgrounds. These methods need to be tested, stressed, and further developed. Additionally, we still need background estimation tools for transient sources.
- The tools still need to be stressed to find limitations. The COSI pipeline team has been rapidly developing the cosipy library in preparation for the satellite mission. Our aim is to make this library robust, sustainable, and highly user-friendly. Through more and more user interactions and feedback, we can better learn where the code is working well, and where it breaks down.
- The way in which parameters are configured needs to be refined, and callable scripts need to be added. By callable scripts we are referring to command-line options that will perform common task, such as producing light curves and spectra.
- Methods need to be developed to determine the response for broadened and offset line emission. These methods should utilize the baseline response files (e.g. 511 keV, Aluminum-26, Iron-60, etc.), and allow for analyzing any arbitrary line emission.
If you make use of any of the data products from the COSI Data Challenges in a publication, please provide a link to this page and cite Zoglauer+23 and Martinez-Castellanos+23.