donovanr

```
def logC_taylor(x, eps=1e-7, taylor_center=0.5, taylor_radius=0.05):

    eps = torch.tensor(eps).type_as(x)
    taylor_center = torch.tensor(taylor_center).type_as(x)
    taylor_radius = torch.tensor(taylor_radius).type_as(x)

    # singular at zero and one, so regularize
    mask = x == 0
    x[mask] = x[mask] + eps

donovanr

donovanr

How to use sshfs to mount qmaster+isilon (or any other ssh-able machine)

Get FUSE and SSHFS

• Download FUSE for Mac and SSHFS from here and install them in that order.

Create a mount point

• Create a mountpoint for e.g. qmaster in /mnt and chown it to you:
  • sudo mkdir -p /mnt/qmaster
  • sudo chown <username> /mnt/qmaster
• Symlink it to the /Volumes diectory (creating it directly in /Volumes causes trouble with unmounting):

donovanr

Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py", line 1039, in _do_call
    return fn(*args)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/client/session.py", line 1021, in _run_fn
    status, run_metadata)
  File "/usr/lib/python3.5/contextlib.py", line 66, in __exit__
    next(self.gen)
  File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/errors_impl.py", line 466, in raise_exception_on_not_ok_status
    pywrap_tensorflow.TF_GetCode(status))
tensorflow.python.framework.errors_impl.NotFoundError: Unsuccessful TensorSliceReader constructor: Failed to find any matching files for ./checkpoints/BEGAN_64_64_0001.tfmod

donovanr

Port forwarding using ProxyJump and ssh tunnels

These instructions assume an ssh client at least as recent as OpenSSH 7.3, otherwise the ProxyJump will fail.

• set up a new host in your ~/.ssh/config file (change rorydm to your own username):

Host tartarus
    HostName n70
    ProxyJump qmaster
    User rorydm
    LocalForward 8888 localhost:9999

donovanr

# install matching nvidia drivers on AWS
# from https://stradajl.readthedocs.io/en/latest/tutorial/
# really should rebuild genomics container on top of nvidia-docker
# ----------------------------------------------------------------

# First update the system and install build-essential:

sudo apt-get update && sudo apt-get upgrade
sudo apt-get install build-essential

donovanr

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [

donovanr

# clone repo
git clone https://github.com/andersx/fragbuilder

# add to path, wherever you put it
export PYTHONPATH="${HOME}/git-repos/fragbuilder":$PYTHONPATH

# install openbabel dependency
git clone git://github.com/openbabel/openbabel.git
cd openbabel
mkdir build

donovanr

install anaconda

add channel to conda config:
conda config --add channels http://conda.binstar.org/omnia

conda install openmm
conda install openmmtools

donovanr

How to install the MMTK python package

On Mac (10.10)

install anaconda
create an environment with anaconda specifically for MMTK (needs numpy<1.9):
- conda create -n mmtk numpy=1.8.2
active that environment
- source activate mmtk
install netcdf and cython: