[boost] GSOC

Subject: [boost] GSOC
From: Flórián Deé (deeflorian_at_[hidden])
Date: 2013-04-22 17:50:00

Personal Details:
 Name: Flórián Deé
 University: Budapest University of Technology and Economics
 Major: Engineering Information Technology
 Degree Program: B.Sc.
 Email: deeflorian_at_[hidden]
 Availability:
  -starting after the exams, best case: June 10, worst case: June 21
  -ending on at the 2nd week at the university: Sept. 15
  -planned time spent on GSoC: 32 to 56 hours per week, as needed
  -other factors: two 3-day long trips - would catch up afterwards with
extra work

Background information:
 During my B.Sc. studies I'll already have finished all courses that are
related to calculus, linear algebra, number theory, combinatorics and
digital technology until summer, so in my opinion, all the educational
background needed is present with the exception of the Signals and Systems
course in next semester, which would cover digital Fourier transforms in
detail. I'd gladly make up for this by reading the needed material.
 As for my programming background, I've fallen in love with C and C++,
which now I consider to be my main languages. I'm currently working on a
project as an outsider member of the dorm's Coder team (a poker framework
for AIs), and I'm also a "consultant" for students with their C/C++
assignments. As a result, I've seen, written and debugged lots of different
(admittedly not so enormous) projects with different coding styles.
 I'm interested in contributing to the Boost Libraries because two of my
friends used a quite a lot of them while researching, and their feedback
was always positive. The C++11 implementations also sounded nice, so adding
to this codebase as a summer project would be an honor.
 I wanted to do a Solar System simulator during secondary school because I
thought that I could calculate the acceleration (and approximate path) of a
pen somehow shot near Earth. Then after the first console-based C++
implementation I realized that doubles aren't sufficiently precise to
determine if an object is near or inside a planet, so I tried to write a
"double double" class - which was too childish to work both code and
mathematicalwise. Since then I know that I'd needed a precise mathematical
library, and that radix-2 is my best friend when calculating on CPU-s, so I
think that this project would be an ideal fit.

Rating of knowledge:
 C++ - 4
 C++ Standard Template Library - 4
 Boost C++ Libraries - 2
 Subversion - 3
Development environments: Eclipse, Visual Studio
Documentation tools: Doxygen

Project proposal, schedule and milestones:
 I'd like to work on the Boost.Multiprecision project described on the wiki
page. As the description there is more than detailed, I wouldn't like to
copy all that text here, but I'd follow the main steps described. I think
that the I/O and the +, - operators should be finished, and the
multiplication already started by the end of July. The *, / operators
should be finished by the end of August at least with the Karatsuba
multiplication, interoperable with Boost.Math, and the FFT multiplication
should also be started by then. The final goal would be FFT multiplication,
with passed correctness tests of wide variety from the beginning of the
project.