akuchling: Sherlock Hemlock (Default)
Barb suggested a mobile app idea. There's a toy that consists of a glass box or tube filled with sand or beads. There are also slightly larger figures buried in the sand -- say, a shark, an anchor, and a ship. You play by shaking the sand until one of the figures comes to the top, and then you can check it off your list. Repeat until you've found all three figures. Her suggestion is a mobile app to play this game: shake your phone until all the stuff comes up.

I think this could probably be done as an JavaScript + canvas app. It could either simulate the physics of sand in detail -- searching for 'sand simulation' turns up papers such as http://robotica.uv.es/~ignacio/papers/EG08/short2054.pdf -- or just fake it like a slot machine, by randomly picking how often a target figure appears and how much of it is visible, and then rendering the screen accordingly.

akuchling: Sherlock Hemlock (Default)
For a book club, I'm currently reading The Hockey Stick and the Climate Wars by Michael E. Mann. Mann is one of the authors of two papers in 1998/99 that attempted to reconstruct historical temperatures for the past millennium or so, and found that the warming in the 20th century is very rapid and has no precedent -- the so-called 'hockey stick' graph.

These papers quickly became the target of many attacks in which people suggested that the data had been incorrectly handled or even fraudently cooked. (It's surely only a coincidence that a fair number of the attackers were from right-wing lobbying groups.)

A while ago I looked at whether I could assist with the Python rewrite of GISTEMP, but it didn't look like there was much I could do productively. The current status of the project is unclear, so the work I attempted turned out to have already been done in a GSoC project.

A new project could be to rewrite the MBH98 code (which seems to be available in the PROXY/ subdirectory from http://www.meteo.psu.edu/holocene/public_html/Mann/research/research.php). It's not clear whether the MBH99 has been released; I don't think so, after browsing through the data directories. There's also a Matlab script for a 2007/8 paper.

I wonder if it would be better to convert the 1998 code -- because it was such a flashpoint, maybe easier reading & experimentation would convince people that the analysis was reasonable -- or the 2007 analysis which is more recent and goes further back.


akuchling: Sherlock Hemlock (Default)

September 2016

2526272829 30 


RSS Atom

Most Popular Tags

Style Credit

Expand Cut Tags

No cut tags
Page generated Jul. 21st, 2017 08:30 pm
Powered by Dreamwidth Studios