michalkonecny / aern Goto Github PK

The problem definition and results are at http://rnc7.loria.fr/comp_result.html.
To be comparable to other tools in 2006, we should get the harder one solved in a few seconds.

The problem definition and results are at http://rnc7.loria.fr/comp_result.html.
To be comparable to other tools in 2006, we should get the harder one solved in a few seconds.

Currently, the test

https://github.com/michalkonecny/aern/blob/master/aern-poly-plot-gtk/demos/testSineCosine.hs

is quite slow to produce not very accurate enclosures of sine and cosine.

In particular, sin(x+1) behaves much worse than sin(x), which should not happen.
Try fix small problems but postpone larger steps, such as trying new representations or a different series than the Taylor series.

Eg when running

bin/simple-ode ivpSpringMass-uv 4 "PlotParam(3)[True,True](-1.2,1.2,-1.2,1.2)" GUI ShrinkWrap 10 200 0 0

the plot should show fairly large uncertainty sets for each sample, but instead only a dot appears, even after a large zoom.

Update the code and the installation instructions so that they work on a fresh install of Haskell Platform 2014.2.0.0.