I have a user who is using the ipython notebook interface in 2016.2-community. When he tries to run an analysis from one of his stored notebooks, he often gets an out of memory error from java. From my reading of the traceback, he's running out of heap while doing routine editing in the notebook interface. (I have a copy of the traceback if anyone cares to read it.) We've already increased his starting heap to 3000MB in the Java options, but this doesn't seem to abolish the problem, only reduce the frequency. Has anyone else had problems like this with the notebook interface, and is there anything we can do to rectify it? The notebooks are only a few megabytes in size (including inline graphics and such), and he's been able to reproduce the error with only a single notebook open, so I don't think that it's the sheer size of the notebook that is the issue, per se.