andrew@0: {\rtf1\ansi\ansicpg1252\cocoartf1038\cocoasubrtf350
andrew@0: {\fonttbl\f0\fswiss\fcharset0 Helvetica;}
andrew@0: {\colortbl;\red255\green255\blue255;}
andrew@1: \paperw11900\paperh16840\margl1440\margr1440\vieww20940\viewh15820\viewkind0
andrew@0: \pard\tx566\tx1133\tx1700\tx2267\tx2834\tx3401\tx3968\tx4535\tx5102\tx5669\tx6236\tx6803\ql\qnatural\pardirnatural
andrew@0: 
andrew@0: \f0\fs24 \cf0 BUGS:\
andrew@0: When prior and posterior etc are split off the screen, it doesn't find the right indices for them. Are we writing the correct prior vector? constrained limits?\
andrew@0: \
andrew@0: \
andrew@0: using best estimate continually - good in the new note on function?\
andrew@0: \
andrew@0: + + + + + + + + + \
andrew@0: \
andrew@0: best estimate locked to map so need to free\
andrew@0: \
andrew@0: need to do relative tempo estimates\
andrew@0: \
andrew@1: \
andrew@1: \
andrew@1: works\
andrew@1: bartok.mid counterexample:\
andrew@1: problem is that the tempo needs to estimate the likelihood confidence that the match is valid\
andrew@1: at the moment all are equally weighted.\
andrew@1: but if we know that the match is (n, k) then prob(loc_n) the phase at note n - gives us probability that the note is correctly identified.\
andrew@1: \
andrew@1: \
andrew@1: //new note\
andrew@1: in comes note\
andrew@1: \
andrew@1: decay speed distribution - so we have uncertainty about tempo\
andrew@1: \
andrew@1: cross update - so we have new prior for position}