File: THINKME

package info (click to toggle)
eprover 2.6%2Bds-3
  • links: PTS, VCS
  • area: main
  • in suites: bookworm
  • size: 21,288 kB
  • sloc: ansic: 331,111; csh: 12,026; python: 10,178; awk: 5,825; makefile: 461; sh: 389
file content (24 lines) | stat: -rw-r--r-- 672 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Think about orientation...these equations have something in common,
but what? How about average term depth of subterms?

f(a, g(b,c))   =    g(f(a,b), f(a,c))
1+2+2+3+3           1+2+2+3+3+3+3

i(f(a,b))      =    f(i(a),i(b))
1+2+3+3             1+2+2+3+3

But this usually orients larger to smaller terms...undesirable!

f(f(a))        = a
1+2+3            1

Average term depth of terminal nodes / max term depth, larger ->
smaller if equal seems to work for these examples. Easier:

Average term depth of terminal nodes / maximum term depth + 1

Now I need an algorithm to generate an ordering for desired
orientations -> Check Joachims LPO algorithm.


Okayti Wonder