Changes from round of JOSS review

doc/sphinx/src/getting-started.rst

@@ -9,16 +9,35 @@ At it's most basic, you can download and compile ``singularity-eos`` with:
 
   git clone --recursive [email protected]:lanl/singularity-eos.git
   cd singularity-eos
-  mkdir bin
-  cd bin
-  cmake -DSINGULARITY_FORCE_SUBMODULE_MODE=ON -DSINGULARITY_USE_FORTRAN=OFF ..
-  make -j
-  make install # optional: install into directory defined via CMAKE_INSTALL_PREFIX
+  cmake -B builddir -S . -DSINGULARITY_FORCE_SUBMODULE_MODE=ON -DSINGULARITY_USE_FORTRAN=OFF -DSINGULARITY_BUILD_EXAMPLES=ON -DSINGULARITY_BUILD_TESTS=ON 
+  cmake --build build --parallel
+  cmake --install build # optional: install into directory defined via CMAKE_INSTALL_PREFIX
 
 This will download ``singularity-eos`` with no optional dependencies and
 compile the capabilities available in that form. For more details, see
 :ref:`our build page <building>`.
 
+You can also run unit tests by calling
+
+.. code-block::
+
+  make test
+
+in the ``build`` directory, which should produce output confirming that
+several tests pass.
+
+Once compiled, you can quickly check your build/install by going to
+``build/example`` and running ``./get_sound_speed_press``. You should be
+able to see the following output:
+
+.. code-block::
+
+  The final values are:
+  rho = 0.987467
+  uu  = 0.00987467
+  P   = 0.0059248
+  cs  = 0.0979796
+
 If the library is in your include and lib paths (or you built it
 in-tree), you can include the ``eos`` part of the library with
 

doc/sphinx/src/models.rst

@@ -698,7 +698,7 @@ If the unitless user parameter :math:`b=\Gamma_0`, the Gruneisen parameter is of
 form where :math:`\Gamma_0 =` constant in compression. These two limitig cases are 
 shown in the figure below.
 
-.. image:: ../SteinbergGammarho.pdf
+.. image:: ../SteinbergGammarho.png
   :width: 500
   :alt: Figure: Demonstration of how the parameter b interpolated between two common approximations for Gamma
 
@@ -978,7 +978,7 @@ The first omitted term in the expansion inside the square brackets is :math:`\Ga
 in fact even better than the common approximation of replacing the full temperature on the Hugoniot with the temperature on the 
 isentrope, that is, the first term :math:`T_0 e^{\Gamma(\rho_0) \eta}`.
 
-.. image:: ../ApproxForTH.pdf
+.. image:: ../ApproxForTH.png
   :width: 500
   :alt: Figure: Different approximations for the temperature on the Hugoniot.
 
@@ -1018,7 +1018,7 @@ By expanding the MGUsup Hugoniot pressure into a power series in :math:`\eta` we
 
 In the figure below we have used :math:`M=20` with these coefficients and show how the divergence in the MGUsup pressure at :math:`\eta = \frac{1}{s}` is avoided in the PowerMG, making it more suitable for modeling high pressures.
 
-.. image:: ../PMGvsMGUsupPress.pdf
+.. image:: ../PMGvsMGUsupPress.png
   :width: 500
   :alt: Figure: Comparing Hugoniot pressure for PowerMG and MGUsup
 

doc/sphinx/src/python.rst

@@ -7,6 +7,10 @@ Singularity EOS provides Python bindings which can be enabled with the CMake
 ``SINGULARITY_BUILD_PYTHON`` option. They provide a 1:1 mapping of the C++ EOS
 types and give access to both scalar and vector functions.
 
+Where you build or install your python bindings to must be included in
+your python module search path. For example, if you do not install,
+this may be ``singularity-eos/build/python``.
+
 .. note::
 
    At this time, all Python bindings are host-only.

doc/sphinx/src/using-kpt.rst

@@ -183,7 +183,7 @@ transportation rate from/to phases ``fromto[k]``, with ``k`` a phase transition
 in ``fromto``, "ij", is composed from the ``gibbs`` index of the "from" phase, :math:`i`, and the ``gibbs`` index of the "to" phase, 
 :math:`j`, as :math:`i*10+j`, and with a single digit integer, "x", interpreted as "0x".
 
-.. image:: ../GibbsOrder.pdf
+.. image:: ../GibbsOrder.png
   :width: 500
   :alt: Figure: How the phase transition index used in several arrays relate to the phase index in the  gibbsorder array.
 

joss-paper/paper.md

@@ -217,7 +217,7 @@ knowledge to compute them as performantly as possible.
 
 Accelerators present new challenges to standard object-oriented
 programming. In particular, not all compiler stacks (such as Sycl
-[@SYCL] or OpenMP Target Offload [@chandra2001parallel])
+[@SYCL] or OpenMP [@chandra2001parallel] Target Offload [@openmp4])
 support relocatable device code, which is required for standard C++
 polymorphism. Even in programming models, such as CUDA [@cuda],
 which do support relocatable device code, polymorphism can be slower

joss-paper/refs.bib

@@ -126,6 +126,14 @@ @book{chandra2001parallel
   publisher={Morgan kaufmann}
 } 
 
+@misc{openmp4,
+    author = {{OpenMP Architecture Review Board}},
+    title = {{OpenMP} Application Program Interface Version 4.0},
+    month = {July},
+    year = {2013},
+    url = {https://www.openmp.org/wp-content/uploads/OpenMP4.0.0.pdf}
+}
+
 @inproceedings{SYCL,
 author = {Reyes, Ruyman and Brown, Gordon and Burns, Rod and Wong, Michael},
 title = {SYCL 2020: More than Meets the Eye},