Page 208 - HOW TO PROVE IT: A Structured Approach, Second Edition
P. 208

P1: PIG/  P2: OYK/
                   0521861241c04  CB996/Velleman  October 20, 2005  2:54  0 521 86124 1  Char Count= 0






                                   194                        Relations
                                   much in deciding how to prove that b = c. The only other fact we know about
                                   b and c is that they are both smallest elements of B, which means ∀x ∈ B(bRx)
                                   and ∀x ∈ B(cRx). The most promising way to use these statements is to plug
                                   something in for x in each statement. What we plug in should be an element of
                                   B, and we only know of two elements of B at this point, b and c. Plugging in
                                   both of them in both statements, we get bRb, bRc, cRb, and cRc. Of course,
                                   we already knew bRb and cRc, since R is reflexive. But when you see that
                                   bRc and cRb, you should think of antisymmetry. Since R is a partial order, it
                                   is antisymmetric, so from bRc and cRb it follows that b = c.
                                     2. Our first goal is to prove that b is a minimal element of B, which means
                                   ¬∃x ∈ B(xRb ∧ x  = b). Because this is a negative statement, it might help to
                                   reexpress it as an equivalent positive statement:
                                            ¬∃x ∈ B(xRb ∧ x  = b)iff ∀x ∈ B¬(xRb ∧ x  = b)
                                                                iff ∀x ∈ B(¬xRb ∨ x = b)
                                                                iff ∀x ∈ B(xRb → x = b).

                                   Thus, to prove that b is minimal we could let x be an arbitrary element of B,
                                   assume that xRb, and prove x = b.
                                     Once again, it’s a good idea to take stock of what we know at this point about
                                   b and x. We know xRb, and we know that b is the smallest element of B, which
                                   means ∀x ∈ B(bRx). If we apply this last fact to our arbitrary x, then as in part
                                   1 we can use antisymmetry to complete the proof.
                                     We still must prove that b is the only minimal element, and as in part 1
                                   this means ∀c(c is a minimal element of B → b = c). So we let c be arbitrary
                                   and assume that c is a minimal element of B, and we must prove that b =
                                   c. The assumption that c is a minimal element of B means that c ∈ B and
                                   ¬∃x ∈ B(xRc ∧ x  = c), but as before, we can reexpress this last statement in
                                   the equivalent positive form ∀x ∈ B(xRc → x = c). To use this statement we
                                   should plug in something for x, and because our goal is to show that b = c,
                                   plugging in b for x seems like a good idea. This gives us bRc → b = c,soif
                                   only we could show bRc, we could complete the proof by using modus ponens
                                   to conclude that b = c. But we know b is the smallest element of B, so of course
                                   bRc is true.
                                     3. Of course, we start by assuming that R is a total order and b is a minimal
                                   element of B. We must prove that b is the smallest element of B, which means
                                   ∀x ∈ B(bRx), so we let x be an arbitrary element of B and try to prove bRx.
                                     We know from examples we’ve looked at that minimal elements in partial
                                   orders are not always smallest elements, so the assumption that R is a total order
                                   must be crucial. The assumption that R is total means ∀x ∈ A∀y ∈ A(xRy ∨
   203   204   205   206   207   208   209   210   211   212   213