User:IssaRice/Faulty mathematical induction proof example: Difference between revisions

From Machinelearning
No edit summary
 
(2 intermediate revisions by the same user not shown)
Line 14: Line 14:
The problem with the proof above is that we are inconsistently bringing in the hypothesis <math>c > 0</math>. This means that we are "doing induction" but without a fixed predicate <math>P(c)</math>, which makes the proof invalid.
The problem with the proof above is that we are inconsistently bringing in the hypothesis <math>c > 0</math>. This means that we are "doing induction" but without a fixed predicate <math>P(c)</math>, which makes the proof invalid.


When proving the vacuous  
When proving the vacuous case, we say that it is vacuously true. This makes it sound like the predicate is <math>P(c)</math> = "if <math>c > 0</math>, then <math>bc > b</math>". With this P in hand, we indeed have that P(0) is true.
 
But now when we get to the induction case, the induction hypothesis ''isn't'' that <math>bc > b</math>. Instead, the IH is that ''if'' <math>c > 0</math>, ''then'' <math>bc > b</math>. So we can't use <math>bc > b</math> without first checking that <math>c > 0</math>.
 
Now we need to show that if <math>c+1 > 0</math> then <math>b(c+1) > b</math>. Actually, when showing an implication, it's ok to just show the [[wikipedia:Consequent|consequent]]. And in any case, <math>c+1 > 0</math> is true for any natural number <math>c</math>, so we aren't gaining any new information. So we need to show <math>b(c+1) > b</math>. Now, ''if'' <math>c > 0</math>, then we can go through the same steps as in the faulty proof to conclude that <math>b(c+1) > b</math>. What if <math>c=0</math>? In this case, we can't use the induction hypothesis, and what we are trying to show is that <math>b1 > 1</math>, which is nonsense.
 
On the other hand, if we had used the predicate <math>P(c)</math> = "<math>bc > b</math>" then the base case wouldn't have gone through.
</div>
</div>
==See also==
* [[User:IssaRice/An induction proof that does not use the induction hypothesis]]

Latest revision as of 15:04, 31 January 2022

Problem statement

Consider the following "proof":

Proposition. Let be positive integers. Then .

Proof. We fix and induct on . For the base case when , the result is vacuously true. Now suppose inductively that we have the result for . Then for we need . But since . Also, by induction hypothesis. Therefore, . This closes the induction.

This proposition is obviously false, since for we have , not . The problem is to figure out where the induction "proof" above goes wrong.

Diagnosis

The problem with the proof above is that we are inconsistently bringing in the hypothesis . This means that we are "doing induction" but without a fixed predicate , which makes the proof invalid.

When proving the vacuous case, we say that it is vacuously true. This makes it sound like the predicate is = "if , then ". With this P in hand, we indeed have that P(0) is true.

But now when we get to the induction case, the induction hypothesis isn't that . Instead, the IH is that if , then . So we can't use without first checking that .

Now we need to show that if then . Actually, when showing an implication, it's ok to just show the consequent. And in any case, is true for any natural number , so we aren't gaining any new information. So we need to show . Now, if , then we can go through the same steps as in the faulty proof to conclude that . What if ? In this case, we can't use the induction hypothesis, and what we are trying to show is that , which is nonsense.

On the other hand, if we had used the predicate = "" then the base case wouldn't have gone through.

See also