1

Let $G = (V, E)$ be a graph, and let $e \in E$ be a cut edge. Show that the removal of $e$ increases the number of connected components by exactly $1$, i.e.,

$$ c(G - e) - c(G) = 1. $$


My Proof:

Suppose $e = \{u,v\} \in E$ and that it is a cut edge. By definition, a cut edge is an edge whose removal increases the number of connected components.

Thus, removing $e$ from $G$ results in a graph $G - e$ where $u$ and $v$ are no longer connected, leading to the formation of two distinct connected components where there was previously one. This implies that the number of connected components strictly increases, giving:

$$ c(G - e) > c(G). $$

Now, consider the graph $G - e$. If we add back the edge $e$, the two disconnected components merge into a single connected component, decreasing the total count by exactly $1$. That is,

$$ c(G) = c(G - e) - 1. $$

Rearranging, we obtain the desired result:

$$ c(G - e) - c(G) = 1. $$


Is this proof sufficiently rigorous? Can someone make it even more formal? I want to avoid any loose terminology and ensure that the argument is mathematically precise.

Rócherz
  • 4,241
  • The only problematic part of your argument in my opinion is "the two disconnected components merge into a single component". What two components are you referring to? $G - e$ may have many components. Evidently you mean the components containing the two endpoints of $e$, but then how do you know they are disconnected (i.e., that the endpoints are in different components)? Why can you say that there are no other components also merged? These are all answerable, and at your stage, you should address them, not leave them for the reader. – Paul Sinclair Mar 14 '25 at 20:09

2 Answers2

1

I do not see any issue in your solution, or your writing.

I'd solve it in this way:
Let's prove it by contradiction, assume after the removal of $e = (u, v)$, 3 vertices $w_1, w_2, w_3$ got disconnected(no pair of them are connected).
Let $p_1 \colon= \text {shortest path from } w_1 \text{ to } w_3$,
$p_2 \colon= \text {shortest path from } w_2 \text{ to } w_3$ and
$p_3 \colon= \text {shortest path from } w_1 \text{ to } w_2$.

Since these vertices got disconnected because of the removal of $e$ their path in $G$ must pass through $e$, then by the pigeonhole principle, at least 2 of these paths pass through $v$ before $u$. We will show in all of 3 cases, 2 of them will still be connected in $G - e$ which is a contradiction.
if the paths were $p_1$ and $p_2$:
$p_1 =$ $<w_1, \cdots, v, u, \cdots, w_3>$, $p_2 =$ $<w_2, \cdots, v, u, \cdots, w_3>$, and we know path from $w_1$ and $w_2$ to $v$ doesn't pass through $e$ because of the assumption that these are shortest paths. so $w_1$ and $w_2$ are connected to $v$, and because of transitive of connection in undirected graphs, we have $w_1$ and $w_2$ are connected; contradiction.

for $p_1, p_3$, by the same reasoning $w_2$ and $w_3$ are connected to $u$ thus connected which is a contradiction.

for $p_2, p_3$:
for $p_2$ and $p_3$ to pass through $v$, it means that there is an edge from the connected component of $u$ to the connected component of $v$ in $G - e$ which contradicts $e$ being a bridge.

Behrad
  • 482
0

I would likely approach this using a proof by contradiction.

Here is my approach:

Let $G = (V, E)$ be a graph, and let $e \in E$ be a bridge. Assume that removing $e$ does not increase the number of connected components by exactly 1. This means one of the following two cases must hold:

  1. The number of components does not increase at all.
  2. The number of components increases by at least 2.

Case 1: The number of components does not increase.

By definition, a cut edge is an edge whose removal increases the number of connected components. However, in this case, removing $e$ does not change the number of components, which contradicts the assumption that $e$ is a cut edge. Thus, this case is impossible.

Case 2: The number of components increases by at least 2.

This would imply that removing $e$ separates $G$ into at least three connected components. However, since $e$ is an edge, it connects exactly two vertices in $G$, meaning that removing it can only separate one component into at most two new components. Therefore, this contradicts our assumption, and this case is also impossible since $u$ and $v$ have to be in different components.

Since both cases lead to contradictions, our assumption must be false. Therefore, removing a cut edge must increase the number of connected components by exactly 1. enter image description here

  • Your argument is simultaneously too detailed and not detailed enough. (It also reads like something an AI would generate, but my point is that it's not good.) The key point is "However, since $e$ is an edge, it connects exactly two vertices in $G$, meaning that removing it can only separate one component into at most two new components" and this needs more detail. Everything else in the answer is fluff. – Misha Lavrov Mar 13 '25 at 21:57