predicates-1.3.csv

;
; predicates-1.3.csv
;
; This example data file demonstrates the learning of arithmetic
; predicates. An "arithmetic predicate" is an arithmetic expression
; that is true or false; so, for example, x + y > 0  is a predicate.
; These can be joined together using boolean operators.
;
; In this example, the dataset reflects the predicate
; (x > 1.3) AND (y > 0.0) AND g and so the combo program that should be
; learned is
;    and( greater_than_zero( plus($x -1.3)) greater_than_zero($y) $g)
;
; To run, enter: "moses -H it -i predicates.csv -W1 -u pred"
;
; To run, with an example exemplar to get things started:
; moses -H it -i predicates.csv -W1 -u pred -e "or(0<( $1) 0<( $2))"
;
; Note: at this time, MOSES has great difficulty learning the above,
; desired predicate, although it does learn a different expression
; which does accurately reproduce the input table.  The reason for this
; difficulty is that MOSES uses a certain binary prepresentation for 
; floating-point constants.  Thus, it can easily learn constants such
; as 1.0, 1.5, 1.25, 1.375, 1.125 and so on, but has difficulty with
; fractions that do not have a power-of-two denominator. Likewise, it
; represents integers in base two: so, for example, 13=8+4+1 requires
; three terms, and thus is more difficult to learn that 5=4+1 which
; has only two terms. At this time, this is a "feature/bug", and its not
; obvious how (or whether) to alter the algorithm to work around this.
;
x	y	g	pred
2.0	1.1	1	1
1.1	1.1	1	0
0.9	2.0	1	0
0.8	-0.1	0	0
1.1	-0.1	1	0
6.6	-0.8	0	0
5.8	0.1	1	1
15.9	4.1	1	1
4.6	0.3	1	1
3.7	0.9	1	1
2.2	1.1	1	1
1.6	1.6	1	1
1.2	2.8	1	0
0.7	3.0	1	0
0.4	6.0	0	0
0.1	8.0	1	0
-0.3	11.0	0	0
-0.6	12.0	1	0
-2.0	23.0	0	0
-9.1	3.0	0	0
-49.1	9.0	0	0
-1.8	4.0	1	0
-0.1	1.0	0	0
0.9	0.2	1	0
1.01	0.1	1	0
1.29	0.05	0	0
1.3	0.05	1	0
1.31	0.05	1	1
1.9	0.01	1	1
2.1	-0.01	1	0
3.1	-0.1	0	0
4.1	-0.6	1	0

# Make sure $g plays a role ... 
15.2	0.1	1	1
5.5	0.1	1	1
4.1	0.4	0	0
3.2	0.6	0	0
2.9	1.2	0	0
1.6	1.5	0	0

# Make sure -$y doesn't get used wrong...
-15.5	3.1	1	0
-5.5	0.1	1	0
-4.1	8.4	1	0
-3.2	0.6	1	0
-2.9	21.2	1	0
-1.6	1.5	1	0

# Make sure that a solution with $x * $y isn't explored 
5.8	19.1	1	1
15.9	0.1	1	1
4.6	20.3	1	1
3.7	15.9	1	1
2.2	72.1	1	1
1.6	31.6	1	1
1.0	12.9	1	0
1.2	32.1	1	0
0.6	32.6	1	0
1.1	1.9	1	0
1.2	3.1	1	0
0.45	2.6	1	0

# Make sure that and(0<(+(-1.5 *(0.5 +(1 $x) $x))) 0<($x)  isnt found.
# The above is actually correct in that its equivalent to 1.3027756...< $x
# but its a polynomial and we'd prefer linear terms...
# 1.305	14.1	1	1
# 1.302	0.1	1	1
# 1.301	23.3	1	1
# 1.3001	25.7	1	1
# 1.2999	2.8	1	0
# 1.2998	1.1	1	0
# 1.298	0.7	1	0
# 1.29	5.3	1	0