
Victor Kores
Jaqueline Mar
CSE 473

README 
Project Part #3
Control Rules and Analysis

1)

One aspect of our PC-CONFIGURATION domain that we were not able to model 
in prodigy had to do with continuous aspects.  We were forced to limit 
our prices to all be multiples of 25.  The goal within-budget needed to 
match a configuration price.  We produced a list of prices within the buget 
of $25 incriments.  The system price the search would find was the price 
of the valid system that unified to the first match in the generated list.
By limiting the prices available to multiples of $25 we were able to have 
a more reasonable search.

Another aspect we were not able to model was the cheapest plan.  In our PC-
CONFIGURATION domain we would consider the cheapest plan a valid configuration
with the minimum cost.  The way we set up the domain was with a budget price 
that prodigy attempted to find a configuration nearest that price.  The 
minimum price could only be found when the problem given had the budget 
reduced to an amount that was near the minimum system price.

b)

The problem of selecting a configuration was trivial.  Adding the requirement 
that that configuration be within a budget constraint was not.  Our first 
attempted goal was to find a solution with any price within the budget.  Our 
final  solution was to produce a list of prices.  The solution seemed like a 
hack but prodigy needed a exact match and our solution provided one at
what seems to be high cost.

c)

A lot of time was spent in TEXT mode.  The traces provided the necessary 
debugging information.  The ANALYZE option was used to examine the final 
state space.

d)

Modification suggestions - What else, run faster.  The machines used did 
not support the need.  There were many hours wasted because of system 
overload.  While this may not be a problem with prodigy it would be nice 
if there was a version that matched the resources available.



Test Case Analysis:

test1: easy problem of finding any valid-configuration with a
	moderate budget.
test2: slightly more difficult.  any valid VGA configuration within
	the budget.
test3: the goal order of test2 is reversed.
test4: more detailed system requirements.
test5: problem 1 with a large budget.


			Number of Nodes
TEST		No-Control		With-Control

test1		878			70
test2		352			336
test3		>1000			336
test4		>1000			728
test5		>1000			>1000


