Removing cython as a dependency

It was only needed for niriss_cython.pyx which (since we currently don't
support NIRISS) could just be a python file instead of a cython file.
This will hopefully resolve our GitHub automated testing

Author: taylorbell57

environment.yml

@@ -1,4 +1,4 @@
-name: eureka
+name: eureka2
 channels:
     - conda-forge
     - defaults
@@ -10,7 +10,6 @@ dependencies:
     - ccdproc
     - celerite # Needed for GP
     - corner
-    - cython
     - dynesty[version='>1.0'] # Lower limit needed for specific arguments
     - emcee[version='>3.0.0'] # Lower limit needed for specific arguments
     - flake8 # Needed for testing
@@ -40,7 +39,7 @@ dependencies:
         - crds
         - exotic-ld
         - image_registration@git+https://github.com/keflavich/image_registration@master # Need GitHub version to avoid np.float issue
-        - jwst==1.10.2
+        - jwst==1.11.4
         - myst-parser # Needed for documentation
         - setuptools_scm # Needed for version number
         - sphinx-rtd-theme # Needed for documentation

environment_pymc3.yml

@@ -1,4 +1,4 @@
-name: eureka_starry
+name: eureka_starry2
 channels:
     - conda-forge
     - defaults
@@ -11,7 +11,6 @@ dependencies:
     - ccdproc
     - celerite # Needed for GP
     - corner
-    - cython
     - dynesty[version='>1.0'] # Lower limit needed for specific arguments
     - emcee[version='>3.0.0'] # Lower limit needed for specific arguments
     - flake8 # Needed for testing
@@ -43,7 +42,7 @@ dependencies:
         - exoplanet
         - exotic-ld
         - image_registration@git+https://github.com/keflavich/image_registration@master # Need GitHub version to avoid np.float issue
-        - jwst==1.10.2
+        - jwst==1.11.4
         - myst-parser # Needed for documentation
         - opencv-python<4.8 # Upper limit needed for numpy<1.22
         - opencv-python-headless<4.8 # Upper limit needed for numpy<1.22

pyproject.toml

@@ -1,5 +1,5 @@
 [build-system]
-requires = ["setuptools>=42", "wheel", "cython", "setuptools_scm[toml]>=6.2"]
+requires = ["setuptools>=42", "wheel", "setuptools_scm[toml]>=6.2"]
 build-backend = "setuptools.build_meta"
 
 [tool.setuptools_scm]

setup.cfg

@@ -23,8 +23,6 @@ package_dir =
 packages = find:
 zip_safe = True
 python_requires= >=3.8
-setup_requires =
-    cython
 install_requires =
     astraeus@git+https://github.com/kevin218/Astraeus@main
     astropy

setup.py

@@ -1,10 +1,6 @@
 #!/usr/bin/env python
 from setuptools import setup
-from Cython.Build import cythonize
 
 if __name__ == "__main__":
 
-    setup(
-        ext_modules=cythonize(
-            ["src/eureka/S3_data_reduction/niriss_cython.pyx"]),
-    )
+    setup()

src/eureka/S3_data_reduction/__init__.py

@@ -5,6 +5,7 @@
 from . import miri
 from . import nircam
 from . import niriss_profiles
+from . import niriss_python
 from . import niriss
 from . import nirspec
 from . import optspex

src/eureka/S3_data_reduction/niriss.py

@@ -21,11 +21,7 @@
 
 from .background import fitbg3
 from .niriss_profiles import *
-
-# some cute cython code
-import pyximport
-pyximport.install()
-from . import niriss_cython
+from . import niriss_python
 
 
 __all__ = ['read', 'simplify_niriss_img', 'image_filtering',
@@ -606,7 +602,7 @@ def construct_guesses(A, B, sig, length=10):
     combos = construct_guesses([0.1,30], [0.1,30], [1,40])
 
     # generates length x length x length number of images and fits to the data
-    img1, sigout1 = niriss_cython.build_image_models(data.median,
+    img1, sigout1 = niriss_python.build_image_models(data.median,
                                                      combos[:,0], combos[:,1], 
                                                      combos[:,2], 
                                                      pos1, pos2)
@@ -619,15 +615,15 @@ def construct_guesses(A, B, sig, length=10):
 
     # generates length x length x length number of images centered around the previous
     #   guess to optimize the image fit
-    img2, sigout2 = niriss_cython.build_image_models(data.median, 
+    img2, sigout2 = niriss_python.build_image_models(data.median, 
                                                      combos[:,0], combos[:,1],
                                                      combos[:,2],
                                                      pos1, pos2)
 
     # creates a 2D image for the first and second orders with the best-fit gaussian
     #    profiles
     final_guess = combos[np.argmin(sigout2)]
-    ord1, ord2, _ = niriss_cython.build_image_models(data.median,
+    ord1, ord2, _ = niriss_python.build_image_models(data.median,
                                                      [final_guess[0]],
                                                      [final_guess[1]],
                                                      [final_guess[2]],
@@ -661,7 +657,7 @@ def residuals(params, data, y1_pos, y2_pos):
         """ Calcualtes residuals for best-fit profile. """
         A, B, sig1 = params
         # Produce the model:   
-        model,_ = niriss_cython.build_image_models(data, [A], [B], [sig1], y1_pos, y2_pos)
+        model,_ = niriss_python.build_image_models(data, [A], [B], [sig1], y1_pos, y2_pos)
         # Calculate residuals:     
         res = (model[0] - data)
         return res.flatten()
@@ -676,7 +672,7 @@ def residuals(params, data, y1_pos, y2_pos):
                                )
 
     # creates the final mask
-    out_img1,out_img2,_= niriss_cython.build_image_models(data.median, 
+    out_img1,out_img2,_= niriss_python.build_image_models(data.median, 
                                                           results.x[0:1], 
                                                           results.x[1:2], 
                                                           results.x[2:3], 

src/eureka/S3_data_reduction/niriss_cython.pyx

@@ -0,0 +1,26 @@
+import numpy as np
+
+def build_image_models(data, A, B, sig, mu1, mu2, return_together=True):
+
+    y = np.transpose(np.full((len(A), data.shape[-1], data.shape[0]),
+                             np.arange(0,data.shape[0],1)), axes=(0,2,1))
+
+    sig = np.full(y.T.shape, sig).T
+
+    A = np.full(y.T.shape, A).T
+    exp = np.exp(-(y-mu1)**2/(2.0*sig**2))
+    gauss = 1.0/(2.0*np.pi*np.sqrt(sig))*exp
+    f1x = mu1*gauss*A
+
+    B = np.full(y.T.shape, B).T
+    exp = np.exp(-(y-mu2)**2/(2.0*sig**2))
+    gauss = 1.0/(2.0*np.pi*np.sqrt(sig))*exp
+    f2x = mu2*gauss*B
+
+    model = f1x+f2x
+    sigma = np.nansum(np.sqrt((model-data)**2.0), axis=(1,2))
+
+    if return_together:
+       return model, sigma
+    else:
+       return f1x, f2x, sigma