The closure of {(x,sin(1/x)):x>0} adds the segment {0}×[−1,1] on the y-axis. Connected as the closure of a connected set, yet no path runs from the segment into the curve — the living example of why connected is weaker than path-connected.
S={(x,sin(1/x)):x>0},S=S∪({0}×[−1,1])
S connected,S not path-connected
Examples & Exercises
Prove completely that S={(x,sin(1/x)):x>0}⊆R2 is connected.
S=f((0,∞)),f(x)=(x,sin(1/x))
xmin=0.05: the curve S alone oscillates infinitely near 0, yet it is one unbroken continuous curve.Show proof
(0,∞) is connected (it is an interval — the same sup argument as for R applies verbatim). The map f(x)=(x,sin(1/x)) is continuous since both coordinate functions are, and its image is exactly S.
The continuous image of a connected set is connected: if f(X)=A∪B were a separation into disjoint nonempty relatively open sets, then X=f−1(A)∪f−1(B) would be a separation of X into disjoint nonempty relatively open sets (A,B nonempty implies their preimages are nonempty), contradicting the connectedness of X.
Hence S=f((0,∞)) is connected. The closure of a connected set is connected, so T=S is connected as well — problem 2 below computes this closure.
Prove completely that S=S∪({0}×[−1,1]): the closure of S is S together with the vertical segment {0}×[−1,1].
S=S∪({0}×[−1,1])
xmin=0.01: zooming closer to 0 shows the sequences (xn,1), (xn,−1) converging onto the endpoints of the red segment.Show proof
⊇: S⊆S is trivial. For (0,y) with ∣y∣≤1, put xn=arcsiny+2πn1 (n≥1); then xn→0+ and sin(1/xn)=sin(arcsiny+2πn)=y. Hence (xn,y)∈S converges to (0,y), so every point of {0}×[−1,1] is a limit point of S.
⊆: let p∈S. If p=(x,y) with x<0, the ball B(p,∣x∣/2) contains no point of S (whose first coordinates are all positive) — impossible. If x>0 and p∈/S, then K=f([x/2,3x/2]) (f from the problem above) is compact as a continuous image of a compact set, hence closed. Points of S meeting B(p,x/2) have first coordinate in (x/2,3x/2), so they lie in K — giving p∈K=K⊆S, a contradiction. Thus x>0 implies p∈S.
Finally, if p=(0,y): for (u,v)∈S, ∣p−(u,v)∣≥∣y−v∣≥∣y∣−1 (since ∣v∣=∣sin∣≤1). If ∣y∣>1, the ball B(p,(∣y∣−1)/2) misses S, contradicting p∈S. Hence ∣y∣≤1, i.e. p∈{0}×[−1,1]. The three cases together give S=S∪({0}×[−1,1]).
Prove completely that T=S is not path-connected: no path in T leads from a point of the segment {0}×[−1,1] to a point of S.
γ(0)∈{0}×[−1,1],γ(1)∈S⟹no continuous γ:[0,1]→T
Try "walking" from the red segment into the curve — the infinite oscillation near 0 blocks every path. The proof captures this obstruction with sequences.Show proof
Assume a continuous γ:[0,1]→T with γ(0)∈{0}×[−1,1] and γ(1)∈S, aiming for a contradiction. Write γ(t)=(x(t),y(t)) and put c=sup{t∈[0,1]:x(t)=0} (nonempty, containing 0). By continuity x(c)=0, and c<1 since x(1)>0. Moreover x(t)>0 for t∈(c,1] (otherwise the supremum would be larger).
For t∈(c,1], γ(t)∈T has positive first coordinate, so γ(t)∈S and y(t)=sin(1/x(t)). Since vn=2πn+π/21→0, we have vn<x(1) for large n. On [c,1], x(c)=0<vn<x(1), so the intermediate value theorem gives tn∈(c,1] with x(tn)=vn. Then y(tn)=sin(2πn+π/2)=1. Every subsequential limit t∗∈[c,1] of (tn) satisfies x(t∗)=limvn=0, forcing t∗=c (as x>0 on (c,1]). Hence tn→c, and by continuity y(c)=limy(tn)=1.
Likewise, with wn=2πn+3π/21, pick sn∈(c,1] with x(sn)=wn; then y(sn)=sin(2πn+3π/2)=−1, and the same argument gives sn→c, so y(c)=−1. This contradicts y(c)=1, so the assumed path cannot exist. T is connected but not path-connected.