Merge pull request #160 from WISDEM/use_conda_pyoptsparse

Use conda from pyoptsparse instead of custom compile

README.md

@@ -57,21 +57,21 @@ The installation instructions below use the environment name, "weis-env," but an
 
 2.  Use conda to install the build dependencies.  Note the differences between Windows and Mac/Linux build systems. Skip to point #3 if you are on the DOE HPC system Eagle
 
-        conda install -y cmake cython control dill git jsonschema make matplotlib-base numpy openmdao openpyxl pandas pip pytest python-benedict pyyaml ruamel_yaml scipy setuptools smt sortedcontainers swig
+        conda install -y cmake cython control dill git jsonschema make matplotlib-base numpy openmdao openpyxl pandas pip pyoptsparse pytest python-benedict pyyaml ruamel_yaml scipy setuptools simpy slycot smt sortedcontainers swig
         conda install -y petsc4py mpi4py                 # (Mac / Linux only)   
         conda install -y compilers                       # (Mac only)   
         conda install -y m2w64-toolchain libpython       # (Windows only)
-        pip install simpy marmot-agents jsonmerge fatpack
+        pip install marmot-agents jsonmerge fatpack
         git clone https://github.com/WISDEM/WEIS.git
         cd WEIS
         git checkout branch_name # (Only if you want to switch git branch, say develop)
         python setup.py develop
 
 3. Instructions specific for DOE HPC system Eagle
 
-        conda install -y cmake cython control dill git jsonschema make matplotlib-base numpy openmdao openpyxl pandas pip pytest python-benedict pyyaml ruamel_yaml scipy setuptools smt sortedcontainers swig
-        conda install -y petsc4py mpi4py  
-        pip install simpy marmot-agents jsonmerge fatpack
+        conda install -y cmake cython control dill git jsonschema make matplotlib-base numpy openmdao openpyxl pandas pip pyoptsparse pytest python-benedict pyyaml ruamel_yaml scipy setuptools simpy slycot smt sortedcontainers swig
+        conda install -y petsc4py mpi4py
+        pip install marmot-agents jsonmerge fatpack
         git clone https://github.com/WISDEM/WEIS.git
         cd WEIS
         git checkout branch_name # (Only if you want to switch git branch, say develop)

WISDEM/.github/workflows/CI_WISDEM.yml

@@ -32,7 +32,7 @@ jobs:
         #shell: pwsh # putting in a shell command makes for compile linking problems later
         # (if you use the shell here, cannot use 'compiler' package)
         run: |
-          conda install -y petsc4py mpi4py compilers nlopt
+          conda install -y petsc4py mpi4py compilers pyoptsparse
 
       # Install dependencies of WISDEM specific to windows
       - name: Add dependencies windows specific
@@ -44,12 +44,26 @@ jobs:
       #- name: Setup tmate session
       #  uses: mxschmitt/action-tmate@v3
 
-      # Install WISDEM
-      - name: Install WISDEM
+      # Install WISDEM with setup
+      - name: Install WISDEM with setup
         shell: pwsh
         run: |
           python setup.py develop
 
+      # Install WISDEM with pip
+      - name: Install WISDEM with pip for Unix
+        if: false == contains( matrix.os, 'windows')
+        shell: pwsh
+        run: |
+          pip install -vv --no-deps -e .
+
+      # Install WISDEM with pip
+      #- name: Install WISDEM with pip for Windows
+      #  if: contains( matrix.os, 'windows')
+      #  shell: pwsh
+      #  run: |
+      #    pip install -vv --no-deps --global-option --compiler=unix .
+
       # Run all tests
       - name: Run tests
         if: contains( matrix.os, 'ubuntu')

WISDEM/README.md

@@ -48,25 +48,19 @@ The installation instructions below use the environment name, "wisdem-env," but
         conda create -y --name wisdem-env python=3.8
         conda activate wisdem-env
 
-2.  In order to directly use the examples in the repository and peek at the code when necessary, we recommend all users install WISDEM in *developer* mode.  This is done by first installing WISDEM as a conda package to easily satisfy all dependencies, but then removing the WISDEM conda package and reinstalling from the Github source code.  Note the differences between Windows and Mac/Linux build systems. For Linux, we recommend using the native compilers (for example, gcc and gfortran in the default GNU suite).
+2.  In order to directly use the examples in the repository and peek at the code when necessary, we recommend all users install WISDEM in *developer / editable* mode using the instructions here.  If you really just want to use WISDEM as a library and lean on the documentation, you can always do `conda install wisdem` and be done.  Note the differences between Windows and Mac/Linux build systems. For Linux, we recommend using the native compilers (for example, gcc and gfortran in the default GNU suite).
 
-        conda install -y wisdem git
-        conda remove -y --force wisdem
-        conda install -y compilers                   # (Mac only)
+        conda install -y cython git jsonschema make matplotlib nlopt numpy openmdao openpyxl pandas pip pyside2 pytest python-benedict pyyaml ruamel_yaml scipy setuptools simpy sortedcontainers swig
+        conda install -y pyoptsparse                 # (Linux only)
+        conda install -y compilers pyoptsparse       # (Mac only)
         conda install -y m2w64-toolchain libpython   # (Windows only)
-        pip install simpy marmot-agents nlopt
+        pip install marmot-agents
         git clone https://github.com/WISDEM/WISDEM.git
         cd WISDEM
         git checkout develop                         # If you want to switch WISDEM branches
         python setup.py develop
 
 
-3. OPTIONAL: Install pyOptSparse, a package that provides a handful of additional optimization solvers and has OpenMDAO support:
-
-        git clone https://github.com/evan-gaertner/pyoptsparse.git
-        pip install -e pyoptsparse
-
-
 **NOTE:** To use WISDEM again after installation is complete, you will always need to activate the conda environment first with `conda activate wisdem-env`
 
 

WISDEM/docs/installation.rst

@@ -13,27 +13,32 @@ Setup and activate the Anaconda environment from a prompt (Anaconda3 Power Shell
 .. code-block:: bash
 
     conda config --add channels conda-forge
-    conda create -y --name wisdem-env python=3.7
+    conda create -y --name wisdem-env python=3.8
     conda activate wisdem-env
 
 Note that any future occasion on which you wish to use WISDEM, you will only have to start with ``conda activate wisdem-env``.
 
 Install WISDEM
 ^^^^^^^^^^^^^^
 
-In order to directly use the examples in the repository and peek at the code when necessary, we recommend all users install WISDEM in *developer* mode.  This is done by first installing WISDEM as a conda package to easily satisfy all dependencies, but then removing the WISDEM conda package and reinstalling from the Github source code.  Note the differences between Windows and Mac build systems.  For Linux, we recommend using the native compilers (for example, gcc and gfortran in the default GNU suite).
+In order to directly use the examples in the repository and peek at the code when necessary, we recommend all users install WISDEM in *developer* mode.  This is done by first installing WISDEM dependencies and then installing WISDEM from the Github source code.  Note the differences between Windows and Mac build systems.  For Linux, we recommend using the native compilers (for example, gcc and gfortran in the default GNU suite).
 
 .. code-block:: bash
 
-    conda install -y wisdem git
-    conda remove --force wisdem
-    pip install simpy marmot-agents nlopt
+    conda install -y cython git jsonschema make matplotlib nlopt numpy openmdao openpyxl pandas pip pyside2 pytest python-benedict pyyaml ruamel_yaml scipy setuptools simpy sortedcontainers swig
+    pip install marmot-agents
+
+For Linux systems:
+
+.. code-block:: bash
+
+    conda install pyoptsparse
 
 For Mac systems:
 
 .. code-block:: bash
 
-    conda install compilers
+    conda install compilers pyoptsparse
 
 For Windows systems:
 
@@ -50,16 +55,6 @@ Finally, for all systems:
     git checkout develop
     pip install -e .
 
-Install pyOptSparse (`Optional`)
-""""""""""""""""""""""""""""""""
-
-`pyOptSparse <https://github.com/mdolab/pyoptsparse>`_ is a package that provides additional optimization solvers with OpenMDAO support:
-
-.. code-block:: bash
-
-    git clone https://github.com/evan-gaertner/pyoptsparse.git
-    pip install -e pyoptsparse
-
 Run Unit Tests
 ^^^^^^^^^^^^^^
 

WISDEM/environment.yml

@@ -26,14 +26,15 @@ dependencies:
   - pyyaml
   - ruamel_yaml
   - scipy
-  - setuptools==58.4.0
+  - setuptools
+  - simpy
   - sortedcontainers
   - swig
   - pip:
-    - simpy
     - marmot-agents
 # Needs to be done outside of environment file:
 #  - m2w64-toolchain # [win]
 #  - libpython # [win]
 #  - compilers # [not win]
 #  - petsc4py # [not win]
+#  - pyoptsparse # [not win]

WISDEM/setup.py

@@ -16,7 +16,7 @@
     extra_compile_args=["-O2", "-fPIC", "-std=c11"],
 )
 pyframeExt = Extension(
-    "wisdem.pyframe3dd._pyframe3dd", 
+    "wisdem.pyframe3dd._pyframe3dd",
     sources=glob.glob(os.path.join("wisdem", "pyframe3dd", "src", "*.c")),
     extra_compile_args=["-O2", "-fPIC", "-std=c11"],
 )
@@ -29,7 +29,7 @@
 # Top-level setup
 setup(
     name="WISDEM",
-    version="3.5.0",
+    version="3.5.1",
     description="Wind-Plant Integrated System Design & Engineering Model",
     long_description="""WISDEM is a Python package for conducting multidisciplinary analysis and
     optimization of wind turbines and plants.  It is built on top of NASA's OpenMDAO library.""",
@@ -43,7 +43,6 @@
         "openmdao>=3.4",
         "openpyxl",
         "pandas",
-        "pyside2",
         "pytest",
         "python-benedict",
         "pyyaml",

WISDEM/wisdem/test/test_optimization_drivers/test_nlopt_driver.py

@@ -10,12 +10,11 @@
 from openmdao.utils.general_utils import run_driver
 from openmdao.test_suite.components.sellar import SellarDerivatives, SellarDerivativesGrouped
 from openmdao.test_suite.groups.sin_fitter import SineFitter
+from wisdem.optimization_drivers.nlopt_driver import NLoptDriver
 from openmdao.test_suite.components.paraboloid import Paraboloid
 from openmdao.test_suite.components.simple_comps import NonSquareArrayComp
 from openmdao.test_suite.components.expl_comp_array import TestExplCompArrayDense
 
-from wisdem.optimization_drivers.nlopt_driver import NLoptDriver
-
 try:
     import nlopt
 except ImportError:
@@ -1001,9 +1000,11 @@ def test_call_final_setup(self):
         with self.assertRaises(RuntimeError) as cm:
             totals = prob.check_totals(method="fd", out_stream=False)
 
-        expected_msg = "Problem: run_model must be called before total derivatives can be checked."
+        # expected_msg = "Problem: run_model must be called before total derivatives can be checked."
+        expected_msg = "run_model must be called before total derivatives can be checked."
 
-        self.assertEqual(expected_msg, str(cm.exception))
+        # self.assertEqual(expected_msg, str(cm.exception))
+        self.assertTrue(str(cm.exception).find(expected_msg) >= 0)
 
     def test_LN_COBYLA_linear_constraint(self):
         # Bug where NLoptDriver tried to compute and cache the constraint derivatives for the

environment.yml

@@ -20,6 +20,7 @@ dependencies:
   - openpyxl
   - pandas
   - pip
+  - pyoptsparse
   - pyside2
   - pytest
   - pytest-cov
@@ -28,6 +29,8 @@ dependencies:
   - ruamel_yaml
   - scipy
   - setuptools
+  - simpy
+  - slycot
   - smt
   - sortedcontainers
   - swig
@@ -36,7 +39,6 @@ dependencies:
     - fatpack
     - jsonmerge
     - marmot-agents
-    - simpy
 # Needs to be done outside of environment file:
 #  - m2w64-toolchain # [win]
 #  - libpython # [win]