Pulling a New Proof from Knuth’s Fixed-Point Printer
32 points by jparise
32 points by jparise
“So. Is there a better program, or a better proof, or a better way to solve the problem?” - possibly these days it would be better to hardcode the answer as a static array. 2^16 is not as large as it used to be ;)
Indeed, Knuth even recognized that in his original exploration. From the post:
Since there are only a small finite number of inputs, Knuth notes that this problem is a counterexample to Djikstra’s remark that “testing can reveal the presence of errors but not their absence.”
These days, that applies to both the proof and potentially the implementation (as you note).