Deadline: 09/30/5:00pm
(1) Open set Property (b) in Theorem 9.3 (page 63): “If U, V are open subsets of Rn, then, U ∩ V is also open.” Let U1, · · · , Um be open sets in Rnwhere m is a natural number. Use mathematical induction and the open set property (b) to prove that U1∩ · · · ∩ Um is also open.
Proof. When n = 1, U1 is open by assumption. Suppose that the statement is true for n = m. Let U1, · · · , Um are open subsets of Rn. By induction hypothesis, U1∩ · · · ∩ Um is open. Let Um+1 be another open subset of Rn. Since
U1∩ · · · ∩ Um∩ Um+1 = (U1∩ · · · ∩ Um) ∩ Um+1,
by open set property (b) and by the fact that both U1 ∩ · · · ∩ Um and Um+1 are open, (U1∩ · · · ∩ Um) ∩ Um+1 is open. Thus U1∩ · · · ∩ Um∩ Um+1 is open. Thus the statement is true for n = m + 1. By mathematical induction, U1∩ · · · ∩ Un is open for any n ∈ N.
(2) Let S = {(x, y) ∈ R2 : xy ≥ 1}. Find int(S). You need to prove that the set int(S)
you found is the right one. Conclude that S is not open by showing S 6= int(S).
Proof. We have proved in class that
int(S) = S \ ∂S.
As long as we show that ∂S = {(x, y) ∈ R2 : xy = 1}, int(S) = {(x, y) : xy > 1}. In this case, int(S) 6= S. Notice that the complement of S is Sc= {(x, y) ∈ R2: xy < 1}.
Let A = {(x, y) ∈ R2 : xy = 1} For p = (x0, y0) ∈ A, x0y0 = 1. Then either x0 > 0 and y0 > 0 or x0< 0 and y0 < 0. We assume that both x0 and y0 are positive. For
> 0, we choose q= (x0+ /2, y0). Then d(p, q) =
2 < .
We find q ∈ B(p, ). Moreover, (x0+
2)y0 = x0y0+
2y0 = 1 +
2y0 > 1.
Hence q ∈ S. We find q ∈ B(p, ) ∩ S. This shows that B(p, ) ∩ S is nonempty for any > 0.
Let us take q0 = (x0− /2, y0). Then d(p, q0) =
2 < .
Hence q0 ∈ B(p, ). Moreover, (x0−
2)y0 = x0y0−
2y0 = 1 −
2y0 < 1.
Thus q0∈ Sc. We see that q0∈ B(p, ) ∩ Sc. Hence B(p, ) ∩ Sc is nonempty for any
> 0.
We conclude that p is a boundary point of S when x0, y0 are both positive. When x0, y0 are both negative, we can apply a similar method to show p is a boundary point of S. Thus A ⊂ ∂S.
1
Let p ∈ ∂S. Let us show that p ∈ A. Let U1 = {(x, y) : xy > 1} and U2 = Sc= {(x, y) : xy < 1}. In homework one, we have shown that U1 is an open subset of R2. In fact, we can also show that U2 is open. By definition, R2 = U1∪ A ∪ U2. If p ∈ U1, by openess of U1, we can find > 0 so that B(p, ) ⊂ U1 ⊂ S. (B(p, ) ∩ Sc is empty.) Then p cannot be a boundary point of S. Similarly, if p ∈ U2, then p is never a boundary point of S. Thus the only possibility is that p ∈ A. We conclude that A = ∂S.
(3) True or false. If the statement is true, prove it; if not, give a counterexample.
(a) If A and B are subsets of Rn, is int(A) ⊂ int(B) when A ⊂ B?
Proof. The statement is true.
Let p be an interior point of A. There exists > 0 so that B(p, ) ⊂ A. Since A is a subset of B, B(p, ) ⊂ B. Thus p is an interior point of B. We find
int(A) ⊂ int(B).
(b) Is it true that int(A) ∩ int(B) = int(A ∩ B)?
Proof. The statement is true.
Let p ∈ int(A) ∩ int(B). Then p is an interior point of A and also an interior point of B. There exist 1, 2> 0 such that B(p, 1) ⊂ A and B(p, 2) ⊂ B. Take
= min{1, 2}. Then B(p, ) ⊂ B(p, i) for i = 1, 2. Thus B(p, ) ⊂ B(p, 1) ∩ B(p, 1) ⊂ A ∩ B.
Thus p is an interior point of A ∩ B; p ∈ int(A ∩ B). Thus int(A) ∩ int(B) ⊂ int(A ∩ B)
If p ∈ int(A ∩ B), then we can choose > 0 so that B(p, ) ⊂ A ∩ B. Thus B(p, ) ⊂ A and B(p, ) ⊂ B. We see that p is an interior point of A and also an interior point of B. We conclude that p ∈ int(A) ∩ int(B). Thus
int(A ∩ B) ⊂ int(A) ∩ int(B).
We conclude that int(A) ∩ int(B) = int(A ∩ B).
(4) Sketch D1 and D2and find the boundary sets of D1 and D2. You need to prove that
∂D1 and ∂D2 you found is correct.
(a) D1 = {(x, y) ∈ R2 : x2− y2> 1}.
Proof. Let U1 = D1 and U2= {(x, y) : x2− y2 < 1} and A = {(x, y) : x2− y2= 1}. Then R2 = U1∪ A ∪ U2 and D1c= U1∪ A. Let us prove that ∂D1 = A.
Step I: Show that U1 and U2 are open.
We leave it to the readers.
Step II: Show that A ⊂ ∂D1.
Let p = (x0, y0) ∈ A. Then x20 − y02 = 1. Either x0 > 0 or x0 < 0. We assume that x0 > 0. In this case, x0 = p1 + y02 > 1. For any > 0, we choose q = (x0+ /2, y0).Then
d(p, q) = 2 < .
Hence q both belong to B(p, ). By > 0 and x0 > 0, (x0+
2)2− y20 = x20+ x0 +2
4 − y20 = 1 + x0 +2 4 > 1.
We find that q ∈ D1. Hence q ∈ B(p, ) ∩ D1. We see that B(p, ) ∩ D1 is nonempty for any > 0.
If 0 < < 1, we take q0 = (x0 − /2, y0). Then d(p, q0) = /2 < . Hence q0∈ B(p, ). By x0> 1,
(x0−
2)2− y02= 1 − x0 +2
4 < 1 − +2
4 = (1 − 2)2 < 1.
We find q0∈ Dc1Hence q0 ∈ B(p, )∩D1cfor 0 < < 1. We see that B(p, )∩Dc1is nonempty for 0 < < 1. If > 1, we take q0= (x0−1, y0). Then d(p, q0) = 1 < .
By x0> 1,
(x0− 1)2− y02= 2 − 2x0 = 2(1 − x0) < 0 ≤ 1.
Then q0∈ B(p, ) ∩ Dc1. Thus B(p, ) ∩ D1cis nonempty for > 1. By the previous result, B(p, ) ∩ D1c is nonempty for any > 0. We find that p ∈ ∂D1. Hence
A ⊂ ∂D1.
Step III: Show that ∂D1⊂ A by openess of U1 and U2.
Let p ∈ ∂D1. Then p ∈ U1 or p ∈ A or p ∈ U2. If p ∈ U1 or p ∈ U2, by openess of U1 and U2, we can show that p is never a boundary point of D1. (The argument is similar as those in the above exercises.) The only possibility is p ∈ A. As a consequence,
∂D1⊂ A.
We conclude that ∂D1= A.
(b) D2 = {(x, y) ∈ R2 : x ≤ y}.
Proof. Let U1 = {x > y} and A = {x = y} and U2 = {x < y}. Then D2c= U2. By homework one, we know U1 and U2 are open. Claim ∂D2= A. Hence points of U1 and U2 are never the boundary points of D2 due to the openess of U1 and U2. We see that
∂D2⊂ A.
Let p = (x0, y0) ∈ A. Then x0 = y0. For any > 0, take q= (x0+ /2, y0) and q0= (x0− /2, y0). Since > 0,
x0+
2 = y0+
2 > y0, x0−
2 = y0− 2 < y0. We see that q∈ U1 while q0∈ U2. Moreover,
d(p, q) = d(p, q0) = 2 < .
We find that q and q both belong to B(p, ). Thus q ∈ B(p, ) ∩ D2, q0∈ B(p, ) ∩ D2c.
Thus B(p, )∩D2 and B(p, )∩Dc2are both nonempty for any > 0. We conclude that p is a boundary point of D2. Hence
A ⊂ ∂D2.
We conclude that A = ∂D2.
(5) Let D = {(x, y) ∈ R2 : x2+ 4y2 ≤ 4} and define f : D → R by f (x, y) =p
4 − x2− 4y2. (a) Find ∂D and int(D).
Proof. Using similar technique as above, one can show that ∂D = {x2+4y2= 4}
and hence int(D) = D \ ∂D = {x2+ 4y2< 4}. The details are left to the reader.
(b) The point p(0, 1) belongs to both of D and ∂D. Is it make sense to you to
define/compute fx(0, 1) and fy(0, 1)? What happen when we compute f (0, 1 + h) − f (0, 1)
h
for h > 0? (We have discussed several examples in class. Please do one by yourself).
Remark: The function f (x, y) is undefined at (0, 1+h) for h > 0. It is nonsense to talk about f (0, 1 + h) and about f (0, 1 + h) − f (0, 1)
h .
(c) Let U be the subset of D consisting of points (x, y) so that fx(x, y) and fy(x, y) both exist. In other words,
U = {(x, y) ∈ D : fx(x, y) and fy(x, y) both exist.}.
Find U. Is U open? If not, find int(U ).
Proof. U = int(D) = {x2+ 4y2 < 4} and U is open. Moreover,
fx= −x
p4 − x2− 4y2, fy = −4y p4 − x2− 4y2.
The proof is given as follows. Let p = (x0, y0). Then p is an interior point of D.
We can find > 0 so that B(p, ) ⊂ D. For |t| < , we set F (t) = f (x0+ t, y0), G(t) = f (x0, y0+ t).
For |t| < , F (t) =
q
4 − 4y02− (x0+ t)2, G(t) = q
4 − x20− (y0+ t)2.
To show that fx(p) and fy(p) exist, it is equivalent to show that F and G are differentiable at 0. In this case, fx(p) = F0(0) and fy(p) = G0(0). We will only prove that F is differentiable at 0 here. The rest is left to the reader. For |t| < ,
F (t) − F (0)
t = −2x0− t
p4 − 4y02− (x0+ t)2+p4 − 4y20− x20
Since limt→0(p4 − 4y02− (x0+ t)2+p4 − 4y02− x20) = 2p4 − 4y02− x20 > 0, we find
limt→0
F (t) − F (0)
t = −x0
p4 − 4y02− x20. This shows that fx(p) exists and equals −x0
p4 − 4y20− x20.
(6) Define a function f : R2→ R by
f (x, y) =
xy(x2− y2)
x2+ y2 if (x, y) ∈ R2\ (0, 0) 0 if (x, y) = (0, 0).
(a) Show that fx(0, 0) = lim
h→0
f (h, 0) − f (0, 0)
h , fy(0, 0) = lim
h→0
f (0, h) − f (0, 0) h
both exist.
Proof. We leave to the readers to check fx(0, 0) = 0 and fy(0, 0) = 0. (b) Let U be the set of all points (x, y) in R2 such that both fx(x, y) and fy(x, y)
exist; find U and evaluate fx(x, y) and fy(x, y) for all (x, y) ∈ U. Is U open?
Proof. We have shown that fx(0, 0) and fy(0, 0) exist. We can also show that for any p 6= (0, 0), fx(p) and fy(p) both exist. Let p = (x0, y0) with x0y0 6= 0.
Then
F (t) = f (x0+ t, y0) = (x0+ t)y0((x0+ t)2− y0)2
(x0+ t)2+ y20 , |t| < .
Here =px20+ y20(Why? Please think about it). Since (x0+t)y0((x0+t)2−y0)2 and (x0+ t)2+ y02 are polynomial functions on R, they are differentiable at 0.
By the quotient rule, F is differentiable at 0 with F0(0) = y0(x40+ 4x20y02− y04)
(x20+ y20)2
Similarly, we can show that fy(p) exists and fy(p) = x0(x40− 4x20y02− y40)
(x20+ y20)2 . We conclude that
fx(x, y) =
0 if (x, y) = (0, 0)
y(x4+ 4x2y2− y4)
(x2+ y2)2 if (x, y) 6= (0, 0).
and
fy(x, y) =
0 if (x, y) = (0, 0)
x(x4− 4x2y2− y4)
(x2+ y2)2 if (x, y) 6= (0, 0).. Hence U = R2 is open.
(c) Show that
fxy(0, 0) = lim
h→0
fx(0, h) − fx(0, 0)
h , fyx(0, 0) = lim
h→0
fy(h, 0) − fy(0.0) h
both exist. Is it true that fxy(0, 0) = fyx(0, 0)?
Using the previous results, we find fx(0, h) = −h and fy(h, 0) = h. Thus fx(0, h) − fx(0, 0)
h = −h − 0
h = −1, fy(h, 0) − fy(0.0)
h = h − 0
h = 1.
We see that fxy(0, 0) = −1 while fyx(0, 0) = 1. Thus fxy(0, 0) 6= fyx(0, 0).