REMINDER: MEETUP: Mutual Mobile, 6PM, Tuesday, Nov. 22
Sukant Hajra
Date: Tuesday, November 22, 2011
Time: 6PM to 8:30PM-ish (however long people want to hack)
Place: Mutual Mobile Offices, 1712 Rio Grande (short walk from campus)
Contact: 512.905.4771 (Sukant's cell phone, in case it helps)
Mutual Mobile moved to the building across the street (from 1709 to 1712).
The building is large, but somewhat nondescript. There's a large staircase
outside that goes up right to the room we meet in. At the base of the
staircase, there's a small metal plaque that reads "Chief Architect" something
or another (I'll figure out exactly what it says Tuesday).
Ignore the sign that says "no entrance". If anything, it's an indication
you're at the exact room we meet at, and that's where we let people in. But
I'll try to put up another sign that's more clear a few minutes earlier to the
meetup.
The calendar [1] and Upcoming Events wiki page [2] are up-to-date.
[1] http://code.google.com/p/atx-sw-mentors/wiki/Calendar
[2] http://code.google.com/p/atx-sw-mentors/wiki/UpcomingEvents
Meeting Agenda
--------------
I've got us slated to continue working on the chemistry balancing program [3].
Last time, Calvin pointed out that the problem really needs integer solutions.
I got to digging a little, and I'm pretty sure this problem is Integer Linear
Programming, which is NP-Hard. If someone feels differently, please let me
know. It's interesting, though -- the problem seemed so benign when I first
read it.
Calvin has a partial solution, which is to solve the problem with inexact
arithmetic (floating point) and then try to reconstruct integers from there.
Clearly, this is subject to problems with rounding errors, but it's certainly a
fair first attempt and may work well for many non-pathological equations.
There /are/ off-the-shelf ILP solvers like lp_solve [4], which is what I'm
planning on trying to use to solve this problem. Hand-rolling an ILP solver is
a neat algorithmic study, but that's slightly more academic than ASM's charter.
I'd like to focus student's attention on other aspects of software development.
[3] http://www.personal.psu.edu/jzl157/balance.htm
[4] http://lpsolve.sourceforge.net/
Tentative for Nov. 29
---------------------
One week from next Tuesday, Lane's volunteered for a public code review of his
work on the balancing problem. Others are invited to participate too.
Comments will be completely constructive. You can't get far in software
without having your code reviewed by someone else, so it's good to overcome any
reticence. And after you're done, you'll realize how much productivity a good
code review can engender.
All of this needs to be sorted out tomorrow, though. Michael and Lane have a
room, but we might want to get a few emails/fliers up too. I'll post back when
everything is concrete.
See everyone Tuesday,
Sukant