Sie sind auf Seite 1von 4

NOTES ON WRITING PROOFS

JEFFREY STOPPLE

Youve already had three quarters of proof-based mathematics: Math 8 Math 108A, and Math 117. In theory you are good at writing proofs; in practice you might still be a little wobbly. These notes are intended to help you get up to speed quickly. Review them before class, and again before you turn in the rst few homework assignments. If you are making these mistakes, I will refer to them explicitly by the section numbers below when I grade your homework. 1. FATAL ERRORS 1.1. Propositions v. objects. A common mistake for beginners is to confuse mathematical objects with true or false statements about these objects. A statement with a true/false value is referred to as a PROPOSITION . For example, P implies Qis nonsense if P is a set. A set is neither true nor false. Impliesconnects two propositions. This is very common with beginners. Conversely x is an element of P makes no sense if P is a proposition such as A B C, because neither x is an element of true nor x is an element of false have any meaning. This is the most serious mistake on this list, such statements are literally nonsensical. To help you decode such things, observe that a proposition is a sentence, so it has a verb. The verb may be symbolic: =, , and all play the role of verbs. 1.2. Assuming the conclusion. In trying to prove P Q, if at some point you write Suppose Q is trueor anything like that, youve assumed what youre trying to prove. This is Sudden Death; I get to stop reading at this point and you get no partial credit. Do not do this under any circumstances. 1.3. True statements v. proofs. Just because the statements in your proof are true does not make them a correct proof. Math 118A is dicult is a true statement; it does not make your proof correct. More subtle is the following. In trying to prove something of the form P Q, it is a fact that P Q is a true statement (or I would

JEFFREY STOPPLE

not assign you to prove it). Equivalently, Because the hypothesis P is true, the conclusion Q is true is in fact a true statement in the abstract, but certainly not a proof. 1.4. Proof by logorrhea. The Oxford English Dictionary denes logorrhea as an excessive ow of words. It is tempting to think if you write more your proof will be better. In fact, correct proofs in this class are likely to be just a couple of lines. If instead it is a couple of pages it is almost certainly wrong. 1.5. Failure to end with the conclusion. When trying to prove P Q, your proof should end with Therefore, Q in some form or other. If you end with a statement like Therefore, R with some completely unrelated proposition R, it is very likely wrong. On the same note, when you state Therefore Q you need to stop writing. If you continue on and furthermore. . . with more justication of why Q is true, your proof is wrong. You must reach the conclusion exactly once, and only at the end of the proof. 1.6. Proof by reduction to tautology. It is very common when trying to prove P Q to begin by writing down Q, and then doing manipulations until an obviously true statement is reached. This is bad for two reasons. First, you cant start by writing down the conclusion and expecting the reader to believe it is true. Second, just because you reach a true statement does not mean your proof is correct; a false hypothesis can be used to prove anything. Below are an incorrect, and a correct proof of the same statement. Note the correct version uses exactly the same ideas. Lemma. Suppose T : V V is a linear map. If v1 , . . . vn are in Ker( T ), then any linear combination c1 v1 + . . . + c n v n Wrong proof: T ( c1 v1 + . . . + c n v n ) = 0 c1 T ( v1 ) + . . . + c n T ( v n ) = 0 c1 0 + . . . + c n 0 = 0 0 = 0. Correct proof: T (c1 v1 + . . . + cn vn ) = c1 T (v1 ) + . . . + cn T (vn ) = c1 0 + . . . + cn 0 = 0. is in Ker( T ).

NOTES ON WRITING PROOFS

1.7. Proof by intimidation. Words like obviously or clearly are a red ag. If it really is obvious or clear you dont need to say so. 2. H IGHLY RECOMMENDED 2.1. Denitions. Make sure you know the relevant denitions. This cant be overemphasized in making the transition from engineering math to proofs. 2.2. Structure. Think about the overall structure of the proof. If it is if and only if, make sure you know it will be two parts and know which part youre working on. Many proofs have a predetermined beginning and end point. Examples include being a subset of, linear independence of some vectors, and limit of a sequence. 2.3. Assuming the hypothesis. In strong contrast to the advice in 1.5 above, it is usually not a good idea to begin the proof of P Q with Suppose P . . . A more helpful thing to do is analyze the statement Q. What properties are involved, and what do you have to do to prove them? (This is a continuation of the advice in 2.2 above.) The hypothesis P is your ace in the hole, to be used when you need it. 2.3.1. In this regard, it is a good idea to mention explicitly when you are using the hypothesis, and worry if you prove the conclusion without using it. 2.4. Neatness counts. It is expected there will be wrong turns and false starts before you gure out the correct solution. You will need to recopy your solution after youve decided its correct - no cross outs or arrows directing the reader around the page. Generally, the easier it is to read, the less time I spend looking at it and the higher the score. 2.5. Proof by contradiction. Despite what you learn in Math 8, in very many cases proof by contradiction is not the best way to proceed, and often a proof by contradiction can be revised to give a more elegant direct proof. Heres an example, the same lemma as above: Lemma. Suppose T : V V is a linear map. If v1 , . . . vn are in Ker( T ), then any linear combination c1 v1 + . . . + c n v n is in Ker( T ).

JEFFREY STOPPLE

Proof by contradiction: Suppose T (c1 v1 + . . . + cn vn ) = 0. Then c1 T (v1 ) + . . . + cn T (vn ) = 0. So for some j, 1 j n, T (v j ) = 0, contradiction. The direct proof above is just the contrapositive.

Das könnte Ihnen auch gefallen