use tensorflow==1.4.0

Author: kaylanb

doc/index.rst

@@ -18,7 +18,7 @@ A Monte Carlo method for adding fake galaxies and stars to images from the Legac
 
 Why the name "obiwan"?
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
-Just as Obi-Wan Kenobi was the "only hope" in Star Wars: Episode IV - A New Hope; our code **obiwan**, by virtue of its Monte Carlo method, is one of the only hopes for removing the systematics in the sample of galaxies we select from the imaging data with our `Legacypipe pipeline <https://github.com/legacysurvey/legacypipe>`_. These are commonly referred to as "imaging systematics" since they are related to image quality, the telescope that took the image, and the bias, bugs, completeness, etc. of the `Legacypipe pipeline <https://github.com/legacysurvey/legacypipe>`_ itself. The Sloan Digital Sky Survey (SDSS), showed that these "imaging systematics" could be removed by measuring correlations between the number of galaxies and image quality (namely, stellar density, seeing, galactic extinction, sky background, and photometric offsets). The situation is far worse for the Legacy Surveys because our images come from three telescopes (SDSS used one), and we take exposures of the same part of the sky, in multiple bands, over timescales of years (SDSS imaged each part of the sky in all bands nearly simultaneously, e.g. over a few minutes). In addition, the cosmological signal we are looking for in the images, about **FILL IN** arcsec, is the size of the field of view of our telescopes' cameras (the SDSS telescope had a field of view of **FILL IN**).
+Just as Obi-Wan Kenobi was the "only hope" in Star Wars: Episode IV - A New Hope; **obiwan**, by virtue of its Monte Carlo method, is possibly the only hope for removing all systematics in the sample of galaxies we select from the imaging data with our `Legacypipe pipeline <https://github.com/legacysurvey/legacypipe>`_. These are commonly referred to as "imaging systematics" since they are related to image quality, the telescope that took the image, and the bias, bugs, completeness, etc. of the `Legacypipe pipeline <https://github.com/legacysurvey/legacypipe>`_ itself. The Sloan Digital Sky Survey (SDSS), showed that these "imaging systematics" could be removed by measuring correlations between the number of galaxies and image quality (namely, stellar density, seeing, galactic extinction, sky background, and photometric offsets). The situation is far worse for the Legacy Surveys because our images come from three telescopes (SDSS used one), and we take exposures of the same part of the sky, in multiple bands, over timescales of years (SDSS imaged each part of the sky in all bands nearly simultaneously, e.g. over a few minutes). In addition, the cosmological signal we are looking for in the images, about **FILL IN** arcsec, is the size of the field of view of our telescopes' cameras (the SDSS telescope had a field of view of **FILL IN**).
 
 It's probably incorrect to describe our Monte Carlo additions of fake galaxies to the imaging data as the "only hope" of identifying imaging systematics, but it should be one of the most robust and self consistent methods for doing so.
 

readthedocs_environment.yml

@@ -36,7 +36,8 @@ dependencies:
   - sphinx_rtd_theme
   - sphinx_bootstrap_theme
   - git+https://github.com/JuliaLang/JuliaDoc.git
-  - tensorflow
+  #- tensorflow
+  - tensorflow==1.4.0
   #- https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.4.0-cp34-cp34m-linux_x86_64.whl
   # mac: --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-1.4.0-py3-none-any.whl
   #- -e git+https://github.com/dstndstn/[email protected]#egg=astrometry