If I may, I’ll run through a similar thing I’ve found for d-CNF clauses.. (conjunctive normal form clauses having ‘d’ elements in each disjunction)

From class, “It’s not always possible to generate a boolean assignment for a d-CNF that yields a true value for the overall clause. However, it is often of interest to generate an assignment that yields a high number of true terms in the clause.”

Their solution is that randomly generating the boolean assignment will have a (2^d – 1)/(2^d) probability of being true anyway, yielding ‘n’ times that for the expected number of true terms in an n-term clause.

My solution is to count up the number of terms in which each boolean assignment occurs, and assign the necessary value to the individual variable that sets most of them true, then repeat this step for the remaining false clauses until you’ve made assignments for all ‘d’ variables. I’m not sure how to prove its correctness, but it seems like a sound argument for maximizing terms… thoughts?

]]>If you feel like going over your proof, let me know. ]]>

Also, for mathematical markup, you may want to check out ASCIIMathML, which is probably the least frivolous use of the language that I’ve seen.

Joe, are you the same Joe Hache who was once president of the CCSS? If so, I’ve got a couple of questions for you.

Also, what the heck are you people doing up at this hour on a Saturday? You’re all nuts.

]]>