Q1: 20 points Q2: 20 points Q3: 20 points Q4: 20 points Q5: 20 points plus 10 extra credit Total: 100 points plus 10 extra credit
Questions 1, 2, and 3 deal with a set T of terriers including two specific terriers Angus (a) and Barney (b). If x is a terrier, the predicate C(x) means "x likes to chase squirrels" and D(x) means "x likes to dig". If x and y are two terriers, SS(x,y) means "x and y are the same size". You are given that SS is an equivalence relation.
If Barney likes to chase squirrels and Angus likes to dig, then Barney both does and does not like to dig.
(C(a) ∨ D(b)) → D(a)
It is not the case that both "if Angus likes to chase squirrels then so does Barney" and "if Barney likes to chase squirrels then so does Angus" are true. (Equivalently: Either Angus or Barney, but not both, likes to chase squirrels.)
∀x:∃y: SS(x,y) ∧ D(y)
If any terrier likes to dig, it does not like to chase squirrels.
∃x:∃y: SS(x,y) ∧ C(x) ∧ ¬C(y)
We can solve this with a sixteen-line truth table, with one line for each of the
sixteen
four settings of the four boolean variables C(a), C(b), D(a), and D(b). We
would find that only one of these settings makes I, II, III, and the two
specifications of V true.
But an equational sequence proof is easier. Since III tells us that either
C(a) or C(b) is true, but not both, let's use Proof By Cases on C(a).
Assume C(a). By Specification from V, D(a) → ¬C(a), so by Modus
Tollens (or Contrapositive followed by Modus Ponens) we get ¬D(a). Again
by Modus Tollens or otherwise from II, we get ¬(C(a) ∨ D(b)), which by
DeMorgan is ¬C(a) ∧ ¬D(b). But ¬C(a) forms a contradiction
with the assumption C(a), so the assumption must be false. We have proved
¬C(a) by Contradiction.
From ¬C(a), C(b) follows from III (by Definition, DeMorgan, and
Definition of Equivalence, III can be rewritten as C(a) ⊕ C(b)).
By Specification on V, we have D(b) → ¬C(b), and C(b) therefore
gives us ¬D(b) by Modus Tollens. Statement I says (C(b) ∧ D(a)) →
0 by Excluded Middle, and thus implies ¬(C(b) ∧ D(a)) which is
equivalent to ¬C(b) ∨ ¬D(a) by DeMorgaon. Since C(b) is true,
D(a) must be false.
We have found that C(b) is true and the other three booleans are false. We
need to check that all our statements hold. I is true vacuously because
C(b) ∧ D(a) is false. II is true vacuously because C(a) ∨ D(b) is false.
III is true vacuously
because C(b) → C(a) is false. The two Specifications of V are also both
true vacuously because D(a) and D(b) are false.
We know from Question 2 that Angus and Barney are two terriers such that one
(Barney) likes to chase squirrels and the other doesn't, but there is no
reason to think that Angus and Barney are the same size. The only way to
guarantee the existence of terriers of the same size is through IV. But how
should we specialize IV?
If we specialize IV to some terrier x, we will get a terrier y of the same
size as x, such that D(y) is true. Then from Specification of V we would get
D(y) → ¬C(y) and thus ¬C(y). This will be enough to prove VI if
our x satisfies C(x). So we want x to be Barney, and the proof goes:
M(a,f) must be true for M to be total, as a does not map to d or e. The other four values must be false for M to be well-defined: M(b,e) and M(b,f) must be false because M(b,d) is true, and M(c,d) and M(c,f) must be false because M(c,e) is true.
Yes, it is both one-to-one and onto. It is one-to-one because each domain value maps to only one range value, and it is onto because each of the three range values is mapped to by at least one domain value.
The simplest thing is to make all five values true. Then we are total because we have M(a,f), M(b,d), and M(c,e), and we are not well-defined because we have both M(b,d) and M(b,e), for example.
The simplest thing is to make all five values false. Then we are not total because there is no y such that M(a,y). But we are well-defined because we never have any domain element mapped to more than one range element -- a, b, and c are mapped to zero, one, and one respectively.
Here are your questions:
T is reflexive: Let x be arbitrary. R(x,x) and S(x,x) are true by
Specification on the reflexivity of R and S respectively. So T(x,x) is true
by the definition of T. Since x was arbitrary, we have proved ∀x:T(x,x).
T is antisymmetric: Let x and y be arbitrary. Assume T(x,y) ∧ T(y,x).
By definition of T and Separation, we have R(x,y) and R(y,x). By Specification
on the antisymmetry of R and Modus Ponens, we have that x=y. We have proved
(T(x,y) & T(y,x)) → (x=y) by direct proof. Since x and y were arbitrary,
we have proved ∀x:∀y:(T(x,y) ∧ T(y,x)) → (x=y).
T is transitive: Let x, y, and z be arbitrary. Assume T(x,y) ∧ T(x,z).
By the definition of T and Separation we get R(x,y), S(x,y), R(y,z), and S(y,z).
By Specification on the transitivity of R to x, y, and z, and Modus Ponens,
we get R(x,z). Similarly from the transitivity of S we get S(x,z). Then
we get R(x,z) ∧ S(x,z) by Conjunction, and this is the definition of T(x,z).
We have shown (T(x,y) ∧ T(y,z)) → T(x,z) by direct proof. Since x,
y, and z were arbitrary, we have proved the transitivity statement by
Generalization.
U is reflexive: Let x be arbitrary. Since we have R(x,x) by Specification
on the relexivity of R, we can get R(x,x) ∨ S(x,x) by Joining, and this is
U(x,x). Since x was arbitrary, we have proved ∀x:U(x,x) by
Generalization.
U need not be antisymmetric: Consider a universe with only two elements a and
b, where only
the reflexive statements, R(a,b), and S(b,a) are true. (Here R and
S are each still antisymmetric and transitive.) Now U(a,b) and U(b,a) are
both true but a is not equal to b.
U need not be transitive: Consider a universe with only three elements a,
b, and c, where only the reflexive statements, R(a,b), and S(b,c) are true.
Then U(a,b) and U(b,c) are both true, but U(a,c) is false.
Last modified 28 September 2007