215 karma · joined January 28, 2011
"He was also drafted into the Wehrmacht to fight on the Eastern front" doesn't seem like he had a lot of choice there...
These often contain bugs that have long been fixed or are incompatible with tools they interface with (e.g. gnuplot/pgf).
I think 12.10 will include texlive 2011 or 2012 (hopefully).
meaning "Permanent usage rights granted at the point of sale are not exclusive to original and may be transferred to a second buyer..."
I'm guessing soon we will only be able to "buy" non permanent usage rights for stuff we buy on steam/random other online distribution platforms.
Some of the "western" bars even have a vpn running so you can browse facebook when connecting to their wifi.
I was just kind of upset about reading the "cost" of his photography experiment/hobby and not even thinking about including the cost of the photo editing software. Especially when it comes to photoshop which is probably one of the most pirated pieces of software due to its price.
I unfortunately know too many hobby photographers who would not even think of paying for this essential tool they probably have been using for many years.
It took me a week or more to download the complete 16 million tweets.
Another problem with the corpus is the fact that (at the time I downloaded the tweets) around 2% of the tweets in the corpus were no longer available from twitter as users deleted their twitter account. The longer you wait, the more tweets are going to be unavailable.
chat works too but the interface is not that great.
This paper [1] is the most recent practical wavelet tree paper I know of.
I used some of the test data on the website to test my implementation [1] but didn't really compare to their other indexes.
Also make sure to always use the chilean mirror as the italian one is outdated (2005).
I recently implemented an FM-Index[1] which supports the following operations:
- count() the number of occurrences of a pattern P of length m in O(m) time.
- locate() determine the locations of all occurrences of a pattern P.
- extract() extract any substring from the Text
- recover the original text
The most interesting part is that the size constructed index depends on the compressibility of the text itself. For English text roughly half the size of the original text.
Construction time and memory requirements, however are not that great.
In the experiment he used patterns of different length on the same text collection. As you can see in the graph, different algorithms perform best for a certain alphabet size.
He describes the text collection as "text corpus taken from wikipedia text dump" so I'm guessing the alphabet size is around 90?
It's also probably not a good thing that all the strings he is searching for are prefixes of the same pattern.
References:
Shift-Or: http://www-igm.univ-mlv.fr/~lecroq/string/node6.html#SECTION...
BNDM: http://www-igm.univ-mlv.fr/~lecroq/string/bndm.html#SECTION0...
BOM: http://www-igm.univ-mlv.fr/~lecroq/string/bom.html#SECTION00...