Merge pull request #109 from sdss/synth

Script to create synthetic MaNGA datacubes

.github/workflows/ci_tests.yml

@@ -19,7 +19,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-latest]
-        python: [3.8, 3.9]
+        python: ['3.9', '3.10', '3.11']
         toxenv: [test-cov, test-astropydev]
     steps:
     - name: Check out repository
@@ -36,11 +36,11 @@ jobs:
         tox -e ${{ matrix.toxenv }}
     - name: Upload coverage to codecov
       if: "contains(matrix.toxenv, '-cov')"
-      uses: codecov/codecov-action@v1
-#      with:
-#        token: ${{ secrets.CODECOV }}
-#        file: ./coverage.xml
-#        fail_ci_if_error: false
+      uses: codecov/codecov-action@v3
+      with:
+        token: ${{ secrets.CODECOV }}
+        file: ./coverage.xml
+        fail_ci_if_error: true
 
   os-tests:
     name: Python ${{ matrix.python }} on ${{ matrix.os }}
@@ -51,13 +51,13 @@ jobs:
       matrix:
 #        os: [windows-latest, macos-latest]
         os: [macos-latest]
-        python: [3.8, 3.9]
+        python: ['3.9', '3.10', '3.11']
         toxenv: [test]
     steps:
     - name: Check out repository
-      uses: actions/checkout@v2
+      uses: actions/checkout@v3
     - name: Set up Python ${{ matrix.python }}
-      uses: actions/setup-python@v2
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.python }}
     - name: Install base dependencies
@@ -70,11 +70,11 @@ jobs:
   codestyle:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
     - name: Python codestyle check
-      uses: actions/setup-python@v2
+      uses: actions/setup-python@v4
       with:
-        python-version: 3.9
+        python-version: '3.11'
     - name: Install base dependencies
       run: |
         python -m pip install --upgrade pip

CHANGES.md

@@ -1,4 +1,11 @@
 
+4.1.0dev
+--------
+
+ - Adds script for creating synthetic MaNGA datacubes for testing.
+ - Change syntax for emission-line modeling files to enable additional tying
+   strategies that make it easier to implement multi-component fitting.
+
 4.0.4 (23 Jun 2022)
 -------------------
 

docs/corrections.rst

@@ -26,17 +26,17 @@ The corrected gas velocity dispersion is:
 
     \sigma_{\rm gas}^2 = \sigma_{\rm gas,obs}^2 - \sigma_{\rm inst}^2
 
-where :math:`\sigma_{\rm gas}` and :math:`\sigma_{\rm inst}` are
+where :math:`\sigma_{\rm gas,obs}` and :math:`\sigma_{\rm inst}` are
 provided in, respectively, the ``EMLINE_GSIGMA`` and
 ``EMLINE_INSTSIGMA`` extensions of the :ref:`datamodel-maps`.
 
 The corrected stellar velocity dispersion is:
 
 .. math::
 
-    \sigma_\ast^2 = \sigma_{\rm obs}^2 - \delta\sigma_{\rm inst}^2
+    \sigma_\ast^2 = \sigma_{\ast,{\rm obs}}^2 - \delta\sigma_{\rm inst}^2
 
-where :math:`\sigma_{\rm obs}` and :math:`\delta\sigma_{\rm inst}` are
+where :math:`\sigma_{\ast,{\rm obs}}` and :math:`\delta\sigma_{\rm inst}` are
 provided in, respectively, the ``STELLAR_SIGMA`` and
 ``STELLAR_SIGMACORR`` extensions of the :ref:`datamodel-maps`.
 

docs/emissionlines.rst

@@ -374,23 +374,31 @@ The columns of the parameter file are:
 | ``action``        | str       | A single character setting how the line should be treated.  See       |
 |                   |           | :ref:`emission-line-modeling-action`.                                 |
 +-------------------+-----------+-----------------------------------------------------------------------+
-| ``tie``           | str[4]    | A sequence of 4 strings that indicate how the line should be tied to  |
-|                   |           | other lines.  The first element specifies the line to which to tie    |
-|                   |           | the line specified in this row, indicated by its line ``index``.  The |
-|                   |           | next three entries indicate how to tie each of the 3 Gaussian         |
-|                   |           | parameters: flux, velocity, and velocity dispersion.  To leave the    |
-|                   |           | parameter untied, this should be ``None``.  To force the parameter to |
-|                   |           | be identical to the parameter for the tied line, use ``=``.  For the  |
-|                   |           | flux, the flux can be specified to be a specific factor of the tied   |
-|                   |           | line; e.g., ``=0.30`` forces the flux of this line to be 0.3 times    |
-|                   |           | the flux of the tied line.  For the velocity and velocity dispersion, |
-|                   |           | a single value can be used to set a the tied parameter to be within   |
-|                   |           | the specified fraction of the tied parameter.  I.e., a value of       |
-|                   |           | ``1.4`` for the velocity dispersion parameter means the velocity      |
-|                   |           | dispersion of this line should be within                              |
-|                   |           | :math:`\sigma_t/1.4 < \sigma < 1.4 \sigma_t`, where :math:`\sigma_t`  |
-|                   |           | is the velocity dispersion of the tied line.  When imposing these     |
-|                   |           | parameter inequalities, `ppxf`_ will require the `cvxopt`_ package.   |
+| ``tie_f``         | str[2]    | A sequence of 2 10-character strings that indicate how the flux of    |
+|                   |           | the line should be tied to another line.  The first element gives the |
+|                   |           | index of the line to tie (see ``index`` above).  The second element   |
+|                   |           | provides the constraint.  Currently fluxes can only be tied by        |
+|                   |           | fixing the line flux ratio, and lines with tied fluxes must also have |
+|                   |           | their velocity and velocity dispersions tied by equality.  For        |
+|                   |           | example, to fix the ratio of the OIII 4959 line to the OIII 5007      |
+|                   |           | line, the entry for the OIII 4959 line should be ``{  14 =0.34 }``,   |
+|                   |           | where 14 is the index number of the OIII 5007 in the file and the     |
+|                   |           | flux in the OIII 4959 line is always 0.34 times the flux in the       |
+|                   |           | OIII 5007 line.                                                       |
++-------------------+-----------+-----------------------------------------------------------------------+
+| ``tie_v``         | str[2]    | A sequence of 2 10-character strings that indicate how the velocity   |
+|                   |           | of the line should be tied to another line.  The first element gives  |
+|                   |           | the index of the line to tie (see ``index`` above).  The second       |
+|                   |           | element provides the constraint.  Velocities can be tied by equality  |
+|                   |           | (using ``=``) or tied by inequality (see below); however, tying by    |
+|                   |           | inequality is not well tested.                                        |
++-------------------+-----------+-----------------------------------------------------------------------+
+| ``tie_s``         | str[2]    | A sequence of 2 10-character strings that indicate how the velocity   |
+|                   |           | dispersion of the line should be tied to another line.  The first     |
+|                   |           | element gives the index of the line to tie (see ``index`` above).     |
+|                   |           | The second element provides the constraint.  Velocity dispersions can |
+|                   |           | can be tied by equality (using ``=``) or tied by inequality (see      |
+|                   |           | below).                                                               |
 +-------------------+-----------+-----------------------------------------------------------------------+
 | ``blueside``      | float[2]  | A two-element vector with the starting and ending wavelength for a    |
 |                   |           | passband to the blue of the primary band.                             |
@@ -409,17 +417,19 @@ and an example file might look like this:
         double restwave;
         char waveref[3];
         char action;
-        char tie[4][10];
+        char tie_f[2][10];
+        char tie_v[2][10];
+        char tie_s[2][10];
         double blueside[2];
         double redside[2];
     } DAPEML;
 
-    DAPEML   2  OII     3727.092   vac  f  {   34   None    = None }  {  3706.3  3716.3 }  {  3738.6  3748.6 }
-    DAPEML   3  OII     3729.875   vac  f  {    2   None    =    = }  {  3706.3  3716.3 }  {  3738.6  3748.6 }
-    DAPEML  23  Hb      4862.6830  vac  f  {   34   None    =  1.4 }  {  4798.9  4838.9 }  {  4885.6  4925.6 }
-    DAPEML  33  NII     6549.86    vac  f  {   35  =0.34    =    = }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
-    DAPEML  34  Ha      6564.608   vac  f  { None   None None None }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
-    DAPEML  35  NII     6585.27    vac  f  {   34   None    = None }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
+    DAPEML   2  OII     3727.092   vac  f  { None   None }  {   34      = }  { None   None }  {  3706.3  3716.3 }  {  3738.6  3748.6 }
+    DAPEML   3  OII     3729.875   vac  f  { None   None }  {    2      = }  {    2      = }  {  3706.3  3716.3 }  {  3738.6  3748.6 }
+    DAPEML  23  Hb      4862.6830  vac  f  { None   None }  {   34      = }  {   34    1.4 }  {  4798.9  4838.9 }  {  4885.6  4925.6 }
+    DAPEML  33  NII     6549.86    vac  f  {   35  =0.34 }  {   35      = }  {   35      = }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
+    DAPEML  34  Ha      6564.608   vac  f  { None   None }  { None   None }  { None   None }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
+    DAPEML  35  NII     6585.27    vac  f  { None   None }  {   34      = }  { None   None }  {  6483.0  6513.0 }  {  6623.0  6653.0 }
 
 .. note::
 
@@ -429,9 +439,12 @@ and an example file might look like this:
       up to the person that writes the two parameter files to make sure
       that is true.
 
-    * The format of this file has changed in version 3.1.0 in a way that removes
-      many of the parameters used by the :class:`~mangadap.proc.elric.Elric`
-      fitter. That class is now deprecated.
+    * Format changes:
+        - version 3.1.0: Many parameters removed that were used by the
+          deprecated :class:`~mangadap.proc.elric.Elric` fitter.
+        - version 4.1.0: Added ability to tie the three parameters to different
+          lines; i.e., velocity can be tied to one line while dispersion is tied
+          to a different one.
 
 .. _emission-line-modeling-action:
 
@@ -442,6 +455,40 @@ Emission-Line "Actions"
 
 .. _emission-line-modeling-mode:
 
+Emission-Line Tying
++++++++++++++++++++
+
+Line tying in the DAP uses the functionality in `ppxf`_ in a limited and
+abstracted way.
+
+**Tying fluxes** effectively means that the lines are put in the same
+emission-line template.  This is why, currently, any lines with tied fluxes must
+also tie their velocity and velocity dispersion.  Also, the DAP currently does
+not allow tying fluxes using inequalities.
+
+**Tying kinematics** can be done with equality or inequality.  For equality, use
+the ``=`` character, as in the example file above.  Unlike the fluxes, the
+kinematics cannot be tied to be, e.g., a specific fraction of the value of the
+tied line.  (I.e., you can't tie the dispersion to be exactly half of the
+dispersion of the tied line).  For inequality, there are a couple of options:
+
+    #. Use ``>N`` or ``<N`` to force the value to be greater or less than the
+       provided fraction of the the value of the tied line.  E.g., to force the
+       dispersion of one component to be at least 1.5 times larger than the tied
+       line, use ``>1.5``.  Using ``>`` or ``<`` is equivalent to ``>1`` and
+       ``<``, respectively.
+
+    #. To bound the value between both upper and lower limits, you must use a
+       *single* fixed fractional bound.  For example, setting the tied value for
+       the dispersion to ``1.4`` means that the best-fitting dispersion must be
+       greater than 1/1.4 and less than 1.4 times the dispersion of the tied
+       line.
+
+.. warning::
+
+    Although line tying has been experimented with for MaNGA data, much of the
+    inequality tying is not well tested.
+
 Emission-Line "Modes"
 +++++++++++++++++++++
 
@@ -454,7 +501,6 @@ Emission-Line "Modes"
 Changing the modeling parameters
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-
 The moment measurements are performed by
 :class:`~mangadap.proc.emissionlinemoments.EmissionLineMoments`; see
 :ref:`emission-line-moments`. A set of parameter files that define a list of

docs/tables/emissionlinepar.rst

@@ -6,7 +6,7 @@ Key            Type           Options                     Default  Description
 ``name``       str            ..                          ..       A name for the line.                                                                                                                                                                                             
 ``restwave``   int, float     ..                          ..       The rest wavelength of the line in angstroms *in vacuum*.                                                                                                                                                        
 ``action``     str            ``i``, ``f``, ``m``, ``s``  ``f``    Describes how the line should be treated.  See :ref:`emission-line-modeling-action`. Default is ``f``.                                                                                                           
-``tie_index``  int            ..                          ..       Index of the line to which parameters for this line are tied.  See :ref:`emission-line-modeling-tying`.                                                                                                          
+``tie_index``  ndarray, list  ..                          ..       Indices the lines to which parameters for this line are tied; one index per (3) Gaussian parameters.  See :ref:`emission-line-modeling-tying`.                                                                   
 ``tie_par``    ndarray, list  ..                          ..       Details of how each model parameter for this line is tied to the reference line.  See :ref:`emission-line-modeling-tying`.                                                                                       
 ``blueside``   ndarray, list  ..                          ..       A two-element vector with the starting and ending wavelength for a bandpass blueward of the emission line, which is used to set the continuum level near the emission line when calculating the equivalent width.
 ``redside``    ndarray, list  ..                          ..       A two-element vector with the starting and ending wavelength for a bandpass redward of the emission line, which is used to set the continuum level near the emission line when calculating the equivalent width. 

examples/fit_one_spec.py

@@ -444,8 +444,6 @@ def main():
               f'{corrected_indices_err[0,i]:12.4f}')
     print('-'*73)
 
-    embed()
-
     print('Elapsed time: {0} seconds'.format(time.perf_counter() - t))
 
 

mangadap/contrib/xjmc.py

@@ -880,11 +880,18 @@ def _fit_iteration(tpl_wave, templates, wave, flux, noise, velscale, start, mome
             # emission lines
             reject_pixels = list(set(pp.goodpixels)
                                     & set(np.arange(len(resid))[pp.gas_bestfit < 1e-6]))
+            if len(reject_pixels) == 0:
+                warnings.warn('Unable to find *good* pixels in the spectrum to use for rejection.'
+                              '  This is likely because a hardcoded threshold could not find '
+                              'pixels that were fit but not part of a fitted emission line.  '
+                              'Please submit a GitHub Issue.  Logging fault and continuing.')
+                fault[i] = True
+                continue
             # - Calculate the 1-sigma confidence interval
             rms = calculate_noise(resid[reject_pixels], width=reject_boxcar)
             # - Reject pixels with > 3-sigma residuals
             model_mask[i,reject_pixels+ps[i]] &= (np.absolute(resid[reject_pixels]) < sigma_rej*rms)
-        
+
             # Reorder the output; sets any omitted components to have
             # the starting values from the original input
             sol, err = _reorder_solution(pp.sol, pp.error, component_map, moments, start=start[i])