What’s New

This section summarises the most important new features a bugfixes in Marvin. For the complete list of changes check the Marvin’s Change Log.

2.6.0 (March 2020)

2.5.0 (September 2020)

  • Adds support for MaNGA release MPL-10 MaNGA data

  • Removes support in the web and API (remote access) for MPLs 4, 5, and 6.

  • Removes support for DAP SPX and VOR10 binning schemes in the web and API access. Only the HYB10 binning scheme is available over the web/API.

  • Note on deprecated support for MPLs and DAP binning schemes in the web/API:

    • MPLs 4, 5 and 6 are still supported via local file mode

    • Alternate DAP binning schemes (SPX, VOR10) are still supported via local file mode

  • MPL-10 datamodel changes, compared to MPL-9:

    • DRP cubes/rss files

      • Extension DISP/PREDISP has been renamed to LSFPOST/LSFPRE in FITS files

      • Marvin MPL-10 Cube/RSS datamodels retain the dispersion/dispersion_prepixel nomenclature for backwards compatibility.

    • DAP maps/modelcube files



      • Added ModelCube (LOGCUBE) LSF extension.

2.4.1 (August 2020)

  • Adds support for switching to the Marvin mirror, magrathea.sdss.org, hosted at Johns Hopkins

2.4.0 (March 2020)

  • Caching of remote Marvin Tools (cached database queries)

  • Improved web performance

    • Adds caching for routes on plate and galaxy pages

    • Improved web spaxel/map handling

  • Added utility functions, and improved handling, of DRPall and DAPall files.

  • Marvin Cube now has a get_available_bintypes method which retrieves the available DAP binning-template schemes available for the given cube.

  • Moving to more semantic versioning (major.minor.patch)

    • major release bumps for breaking changes to API

    • minor release bumps for feature changes and some patches

    • patch release bumps for bug fixes

  • Added support for the Visual Morphology VAC

  • Moved to internal use of tree 3.0.0 and sdss_access 1.0.0

2.3.5 (December 2019)

  • Support for MPL-9 MaNGA data

  • Enhancements to the web spectrum display

    • You can now toggle common lines measured by the DAP (annotations placed using global NSA redshift)

    • You can now toggle between rest-frame and observed wavelength (shifted using global NSA redshift)

    • You can toggle a bad pixel mask. Pixel regions marked as DONOTUSE are now displayed as a red highlighted region.

  • Improved image loading on Image Roulette page

  • “Traditional” spaxel queries have been reimplemented until ongoing improvements can be completed

2.3.4 (December 2019)

Along with plenty of bug fixes, the following things are new

  • Support for DR16, which is the same as DR15.

  • Added support for the following MaNGA VACS

    • GEMA: Returns GEMA table data

    • Galaxy Zoo: Returns Galaxy Zoo morphologies

    • Firefly: Returns stellar population parameters fitted by FIREFLY

    • MaNGA HI: Returns HI summary data and spectra

  • A new Marvin VACs tool to access the full VAC FITS file.

  • Complete support for Windows. With an update to sdss_access, Marvin now fully works on Windows, using curl as the default download service.

  • No more API rate limiting. This has been removed until further notice.

  • New info popups for “DAP Maps” and “Queries” with links to additional information. Click the “DAP Maps” and “Learn More’ labels with the question marks, on the galaxy and query pages, respectively.

  • Improved galaxy web memory - the galaxy web page now remembers when you’ve selected the dynamic MapSpecView, as well your map and bintype selections. Quickly move from galaxy to galaxy without reselecting the same things over and over.

  • Improved DataModel documentation

2.3.2 (February 2019)

Support for MPL-8. MPL-8 includes several changes to the DAP Datamodel. Please see the MPL-8 DataModel and the MPL-8 TRM page (SDSS collaboration access required). In summary, the changes include:

  • The stellar template is now MILESHC-MILESHC, indicating the template used for the stellar kinematics fitting and the emission-lines, respectively.

  • The binning types now include SPX, in addition to HYB10 and VOR10. HYB10 is still the default bintype.

  • Changes to ModelCubes

    • Removed emline_base_fit Model

    • New stellar_fit Model

  • Changes to Maps

    • The stellar_sigmacorr now has two channels.

    • New stellar_fom multi-channel property. This property contains figure-of-merit metrics for the stellar kinematics.

    • Moved stellar_cont_fresid and stellar_cont_rchi2 into the stellar_fom property.

    • Three new emission-lines: HeI 3889, NI 5199, NI 5201

    • New emline_ga multi-channel property; the amplitude of the fitted Gaussian

    • New emline_ganr multi-channel property; the amplitude/noise of the fitted Gaussian

    • New emline_fom multi-channel property; full spectrum figure-of-merits for the emission lines

    • New emline_lfom multi-channel property; reduced chi-square figure-of-merits for the emission lines

  • The DRPall file now has two data extensions, MANGA, and MASTAR. The MANGA extension contains all MaNGA-lead galaxy observations. The MASTAR extension contains all mini-bundles and MaStar (APOGEE-lead) targets of observation. Marvin currently does not support MaStar observations.

2.3.0 (December 2018)


Fix link to authentication.

We are excited to introduce Marvin 2.3.0, the first version of Marvin that provides support for public releases of MaNGA data. With this change, that coincides with the release of SDSS DR15, Marvin ceases to be just an internal collaboration tool and becomes available to the whole astronomical and educational communities.

While Marvin now allows unrestricted access to DR15 data, it still supports access to proprietary MaNGA data (MPL-4 to MPL-7). This double access mode makes necessary the implementation of a new authentication framework.

This version brings new and exciting features, such as an improved RSS and a reimplemented getAperture method. We have also restructured (and in many cases rewritten) the online documentation; we hope this new structure will lower the learning curve for new users, and make advance features easily accessible to those who are more proficient. Finally, many (many) bugs have been squashed and we have implemented numerous small improvements. A full list is available in the changelog.

The following subsections describe some of the major changes in detail. As always, full documentation for the new features is available.

Support for public data releases

Reimplemented RSS

The RSS Tool had never been a totally functional Galaxy Tool. In this version we have completely refactored the RSS class and it is now working great! Instances of RSS consist of a list of RSSFiber (basically a Spectrum with additional attributes), one for each IFU fibre and observation. Observing information is available via the obsinfo attribute. See the documentation for further details.

The new RSS class is well tested but given the magnitude of the refactoring we are offering it in beta state. We appreciate your bug reports and any suggestions on how to improve it.

Extracting all the spaxels in an aperture

Early versions of Marvin included a Cube.getAperture method that allowed to extract the spaxels contained in a geometrical aperture. That feature was deemed not science-ready and removed in following releases. In this version we are reintroducing it as a mixin, GetApertureMixIn, that provides the getAperture method to Cubes, Maps, and ModelCubes. The mixin makes heavy use of photutils to define geometric regions (elliptical, circular, rectangular) either in the image frame or using on-sky coordinates. Selecting the spaxels within a circular region around a set of coordinates is now as easy as doing

>>> ap = cube.getAperture((232.546173, 48.6892288), 5, aperture_type='circular')
>>> spaxels = ap.getSpaxels()

Full documentation is available here. As with the RSS class, please double check any result before using it for science publications. We welcome any feedback on how to improve this feature.

A new tool for handling MaNGA images

A new Marvin Image Tool is now available for interacting with MaNGA optical images. This tool utilizes the Marvin MMA system for easier access to images locally or remotely. The new tool provides plotting as a Matplotlib figure, which includes the image WCS information, as well as options for overlaying individual IFU or sky fibers, and customizing the IFU hexagon. You can also

>>> from marvin.tools import Image
>>> im = Image('8485-1901')

The old image utility functions documented here remain but have now been deprecated by the new Image class. Replacement utility functions are now available, with more information located here.

2.2.6 (August 2018)


This is a critical bugfix release that corrects a problem that could affect your science results. Please update as soon as possible and check whether your analysis has been impacted by this bug.

This version fixes a critical bug when retrieving the spaxels associated with a bin, as well as a problem with the calculation of the inverse variance for deredden datacubes. It also simplifies the library namespace allowing for easier access to the most used Tools.

Critical bugfixes

Spaxels associated with a bin

In version 2.2 we introduced the concept of Bin as a collection of spaxels that belong to the same binning unit. As part of the API, one can use the spaxels attribute to access a list of the spaxels that are included in the bin. The bug now fixed caused a list of incorrect spaxels to be associated with the bin, due to an inversion in the (x, y) order of the spaxels. Before 2.2.6 one would get

>>> cube = Cube('8485-1901')
>>> maps = cube.getMaps('HYB10')
>>> bb = maps[22, 14]
>>> bb.spaxels
[<Marvin Spaxel (x=21, y=13, loaded=False),
 <Marvin Spaxel (x=21, y=14, loaded=False),
 <Marvin Spaxel (x=22, y=13, loaded=False),
 <Marvin Spaxel (x=22, y=14, loaded=False)]

where the x and y values should be

[<Marvin Spaxel (x=13, y=21, loaded=False),
 <Marvin Spaxel (x=14, y=21, loaded=False),
 <Marvin Spaxel (x=13, y=22, loaded=False),
 <Marvin Spaxel (x=14, y=22, loaded=False)]

Inverse variance for deredden datacubes

The DataCube quantity includes a deredden method that applies the reddening correction to the flux and inverse variance in the datacube. The inverse variance associated to the derredden flux had a bug in its calculation and was incorrect in all cases. That has now been fixed. It also fixes the spelling of deredden (😅).

MPL-7 now available

Starting with this release, Marvin provides access to MPL-7, the latest MaNGA Product Launch. Please read the release notes (SDSS collaboration access required) for this version to understand what has changed. MPL-7 is now the default release when Marvin is imported.

In MPL-7 we have made HYB10 the default bintype. Hybrid-binned Maps and ModelCubes use different binning schemes depending on the property measured. Before using HYB10 make sure you read the documentation and understand how to use the data. For HYB10 the Bin class is somehow limited, since it does not allow for different binning schemes depending on the measured quantity. We are planning a major reimplementation of how bins are handled, which we will release with Marvin 2.3.0. In the meantime, please be aware of these limitations when using HYB10.

Simplifying the namespace

Prior to 2.2.6 accessing different Tools classes was inconvenient since one would need to import them independently (e.g., from marvin.tools.cube import Cube, from marvin.tools.maps import Maps, etc.) This version makes access easier by exposing all the Tools from the marvin.tools namespace so that you can now do

import marvin
cube = marvin.tools.Cube('8485-1901')
maps = marvin.tools.Maps('7443-12701')

Passing keyword arguments to Spectrum.plot

Extra arguments passed to Spectrum.plot are now redirected to matplotlib.axes.Axes.plot. This provides extra flexibility for your plots. For instance, you can now set labels for the legend associated with your plot

ax = spectrum.plot(use_std=True, label='flux')
ax.plot(spectrum.wavelength, model_flux, label='model')

Stellar Sigma Correction

For MPL-6, we now raise an explicit error when attempting to apply the correction to stellar_sigma, using the inst_sigma_correction method. The error message now suggests to upgrade to MPL-7 data. For the web display of the stellar_sigma and emline_gsigma maps, we now apply the sigma correction automatically. The corrected map is indicated via Corrected: stellar_sigma map title.

2.2 (January 2018)

Marvin 2.2.0 brings significant improvements in the way you interact with MaNGA data. Try the Jupyter Notebook for a small sample.

  • MPL-6 compatible (SDSS collaboration access required)

  • New DRP, DAP, and Query Datamodels

  • Cubes, Maps, and ModelCubes now use Astropy Quantities, i.e. encapsulating a measurement with its associated parameters (e.g., unit, mask, or inverse variance)

  • Improved Bin class

  • Fuzzy Searching and Tab Completion

  • New access to DAPall data on Maps and ModelCubes

  • Scatter and Histogram Plotting

  • Improved Query Results Handling and Integrated Plotting

  • New MaskBit class

2.1.4 (August 2017)

  • Refactored the Query Page in Marvin Web: Adds more intuitive parameters naming in dropdown. Adds Guided Marvin Query Builder, using Jquery Query Builder. See the Search page section of Web Docs.

  • Adds Galaxy Postage Stamp view of the result set from a Marvin Query in the Web

  • Adds Rate Limiting for the Marvin API. Adopts a limit of 200 requests/min on all routes and 60/min for queries.

  • Adds new query_params object in Marvin Tools for improved navigation and selection of available query parameters. See updated documentation for Queries and Query Params

  • Adds ability for creating custom maps (using custom values and masks) with Marvin Plotting framework. See updated Plotting Tutorial

  • New Sidebar in Marvin Documentation for easier navigation.

  • New Marvin Getting Started Page.

  • New Marvin Exercises for showcasing utilization of Marvin in science workflows

  • Numerous bug fixes. See Changelog for full account of all Github Issues closed.

2.1.3 (May 2017)

  • Slicing in tool objects now behaves as in a Numpy array. That means that cube[i, j] returns the same result as cube.getSpaxel(x=j, y=i, xyorig='lower').

  • Now it is possible to query on absolute magnitude colours from NSA’s elpetro_absmag. Absolute magnitudes are now the default for plotting on the web.

  • The data file for the default colormap for Map.plot() (“linear_Lab”) is now included in pip version of Marvin and does not throw invalid FileNotFoundError if the data file is missing.

  • Query shortcuts are now only applied on full words, to avoid blind replacements. This fixes a bug that made parameters such as elpetro_absmag_r being replaced by elpetro_absmaelpetro_mag_g_r.

  • Refactored Map plotting methods into Utilities.

    • Map plotting now accepts user-defined value, ivar, and mask arrays (e.g., BPT masks).

    • It is possible to create multi-panel map plots.

    • All plotting code no longer overwrites matplotlib rcParams.

    • Map plotting has new default gray/hatching scheme for data quality (in tools and web):

      • gray: spaxels with NOCOV.

      • hatched: spaxels with bad data (UNRELIABLE and DONOTUSE) or S/N below some minimum value.

      • colored: good data.

    • Map plotting no longer masks spaxels near zero velocity contour because by default (in tools and web), there is no minimum signal-to-noise ratio for velocity plots.

  • New tutorials: Plotting Tutorial and Lean Tutorial.

2.1 (February 2017)

  • Marvin is now minimally compliant with Python 3.5+

  • https://sas.sdss.org/marvin now points to Marvin 2 (instead of Marvin 1).

  • The NSA catalog information is now available via Cube.nsa in Marvin Cubes.

  • Marvin marvin-web now has a new Galaxy Properties Display tab with interactive scatter, and box-and-whisker plots.

  • Marvin marvin-web has more python tips for working with Marvin Query Tools objects.

  • Marvin now uses Sentry to catch and send errors.

  • Marvin Maps now include the ability to make and plot a BPT Diagrams diagram.

  • Marvin Maps have updated plotting display and now include a new signal-to-noise (snr) attribute on each map.

  • Check out the visual-guide.

  • Marvin Spaxels now include ra and dec as properties.

  • Streamlined list of query parameters both in the marvin-web and Tools (marvin.tools). Added new parameter ha_to_hb ratio.

  • Marvin has updated the Image Utilities functions for downloading, showing, and locating.

  • New check_marvin utility to provide some basic system checks with regards to Marvin

  • Marvin marvin-web now has a “Provide Feedback” button in the navbar that directly links to a New Issue in Github.

  • See Changelog for more.

2.0 Beta (November 2016)

  • Brand new painless installation (pip install sdss-marvin)

  • New Marvin Tools (Maps, Bin, ModelCube)

  • Pickling of Marvin Tools, Queries, and Results (i.e. local save and restore)

  • DAP Spaxel (“Zonal”) Queries

  • Dynamic DAP Map display in the web, with point-and-click spaxel

  • For MPL-5+, display of model fits in spectrum view in the web

  • Versions simplified from mpl, drp, dap down to release

  • API Authentication

2.0 Alpha (June 2016)

Marvin 2.0 is a complete overhaul of Marvin 1.0, converting Marvin into a full suite of interaction tools.

Marvin 2.0 introduces two new modes of operations, Galaxy Tools and API, to the Marvin environment, and introduces an extensive redesign of the Marvin web app.

The major improvements and additions in this release:

  • Data Access Modes: a new mode based navigation system that allows you to seamlessly interact with MaNGA data no matter where it is.

  • Galaxy Tools: a python package for accessing and interacting with MaNGA data, whether the files are in your computer or they need to be retrieved remotely via the API.

  • API: remotely grab the data you are looking for as JSONs to integrate directly into your local scripts

  • ../query: a tool to harness the full statistical power of the MaNGA data set by querying the Databases.

  • A completely overhauled Web interface, including:

    • A more powerful Search Page with an intuitive pseudo-natural language search capability.

    • A simple and clean Plate and Galaxy detail page.

    • Interactive spectrum selection from the galaxy image.

    • An image roulette if you are feeling lucky.