Dinghua Li voutcn

Installing gcc 4.8 and Linuxbrew on CentOS 6

The GCC distributed with CentOS 6 is 4.4.7, which is pretty outdated. I'd like to use gcc 4.8+. Also, when trying to install Linuxbrew you run into a dependency loop where Homebrew's gcc depends on zlib, which depends on gcc. Here's how I solved the problem.

Note: Requires sudo privileges.

Resources:

http://superuser.com/a/676337/88393: Forum response on using CERN's open Scientific Linux distribution of RHEL's developer toolset.
http://linux.web.cern.ch/linux/devtoolset/: CERN's developer toolset installation instructions.

Writing a program for efficient matrix multiplication is tricky. A naive implementation usually looks like this (I will use a square matrix for simplicity):

for (i = 0; i < n; ++i)
  for (j = 0; j < n; ++i)
    for (k = 0, m[i][j] = 0.; k < n; ++i)
      m[i][j] += a[i][k] * b[k][j];

The problem is the inner loop a[i][k] * b[k][j] where b[k][j] and b[k+1][j] are not adjacent in memory. This leads to frequent cache misses for large matrices. The better implementation is to transpose matrix b. The implementation will look like:

for (i = 0; i &lt; n; ++i) // transpose

	#!/bin/bash

	echo "WARN: This will remove everything from docker: volumes, containers and images. Will you dare? [y/N] "
	read choice

	if [ \( "$choice" == "y" \) -o \( "$choice" == "Y" \) ]
	then
	sudo echo "> sudo rights [OK]"
	sizea=`sudo du -sh /var/lib/docker/aufs`

	"""
	Minimal character-level Vanilla RNN model. Written by Andrej Karpathy (@karpathy)
	BSD License
	"""
	import numpy as np

	# data I/O
	data = open('input.txt', 'r').read() # should be simple plain text file
	chars = list(set(data))
	data_size, vocab_size = len(data), len(chars)

	This document is a partial presentation of the RopeBWT2 pre-print

	http://arxiv.org/abs/1406.0426

	It is the transcript of a presentation made within the Medvedev group at Penn State
	in July 2014. It focuses on illustrating some notions from the methods, and
	illustrating and proving Algorithm 1. While this document does not cover the main
	contribution of the RopeBWT2 paper, I hope it can be helpful towards understanding
	the theoretical foundations that led to Algorithms 2 and 3.