PyCall Issues with MKL

General Usage
1

I want to call a tiny bit of Python code that makes use of some FEniCS code, so I tried adding PyCall and building it against the (conda) Python environment on my system for which I have FEniCS working. However, this ends up not working, and, in fact, anything that relies on PyCall (i.e., PyPlot) then seems to break:

julia> using PyPlot
INTEL MKL ERROR: dlopen(/Users/gideonsimpson/miniconda3/lib/libmkl_intel_thread.1.dylib, 9): Library not loaded: @rpath/libiomp5.dylib
  Referenced from: /Users/gideonsimpson/miniconda3/lib/libmkl_intel_thread.1.dylib
  Reason: image not found.
Intel MKL FATAL ERROR: Cannot load libmkl_intel_thread.1.dylib.

Anyone have any advice on how to resolve this? I find that if I blow away my .julia folder, and reinstall PyPlot without changing ENV["PYTHON"], I at least can restore PyPlot functionality.

2

As an addendum, the issue seems to be the Numpy installation and how its linked against MKL. FEniCS is a nightmare to install properly, and I’m leery of monkeying with it. Does anyone have a recipe with how, once I have a Python installation working the way I want, to get PyCall to interface with it properly?