12.1 MST practice
a)
After the first step: After the second step:
b)
c)
12.3 Uniqueness of MSTs
a)
We can get MST from Kruskal’s algorithm, and from Prim’s algorithmnot unique MST.excalidraw
⚠ Switch to EXCALIDRAW VIEW in the MORE OPTIONS menu of this document. ⚠ You can decompress Drawing data with the command palette: ‘Decompress current Excalidraw file’. For more info check in plugin settings under ‘Saving’
Excalidraw Data
Text Elements
A
B
C
1
1
1
Link to original
b)
Since is a spanning tree, adding an extra edge creates exactly one cycle: Assume connects and . There exists a path from to through the spanning tree (without using ). If we add to this path, we get a cycle.
We claim this cycle to be . There must be an edge on , because otherwise would be fully contained in , which contradicts the fact that is a spanning tree. Moreover, could not be , because . So and are distinct. Since all weights are distinct and is the minimum weight, we have
c)
We replace with on . We first add to and get a cycle , then we remove another edge on this cycle, such that the remaining structure is still a spanning tree (every vertices are still connected and no more cycles). Now we’ve reduced the total weight of by (not ). This leads to a contradiction to our assumption that is a minimum spanning tree. Therefore, the minimum spanning tree of with weight function is unique.
d)
There is one unique minimum spanning tree , however, the weights are not distinct.Not unique weight of MST.excalidraw
⚠ Switch to EXCALIDRAW VIEW in the MORE OPTIONS menu of this document. ⚠ You can decompress Drawing data with the command palette: ‘Decompress current Excalidraw file’. For more info check in plugin settings under ‘Saving’
Excalidraw Data
Text Elements
A
B
C
1
1
Link to original
12.4 Counting Minimum Spanning Trees With Identical Edge Weights
a)
Contracting an edge just “glues together” its two endpoints. Since is a tree, this operation does not create cycles or disconnect the graph. So the contracted tree is still a spanning tree of the contracted graph . Now we show that is minimal. Suppose we uncontract (separate , and add back the original edges). The total weight increases by exactly .If there were a spanning tree of cheaper than , then uncontracting it would give a spanning tree cheaper than , which contradicts the fact that is minimal.
b)
We partition into weight classes. we define An edge is redundant if it lies in a class of size at least . ( with ) Let . In , the only weight class whose size changes is , which becomes size . Case 1: , then the redundancy drops by 2. because now has size of 1 and does not contribute to the overall redundancy. Case 2: , then the redundancy drops by 1. Therefore, the number of redundant edges in is at most Hence, is -redundant for some
c)
We prove by induction Claim: If is connected and k-redundant, it has at most distinct MSTs. Base: : If , then all edges are pairwise distinct, so MST is distinct (result from 12.3). The number of distinct MST is
Step: we assume if is connected and k-redundant, it has at most distinct MSTs Consider a (k+1)-redundant : we pick a redundant edge We denote the number of distinct MSTs that contains with We denote the number of distinct MSTs that does not contains with
case 1 by (a) the contracted tree is an MST of , so by (b), is at most ((k+1)-1)-redundant. So we have case 2 Consider . Any MST with is still a spanning tree of . Hence we have: we know deleting a redundant edge decreases the number of redundant edges by at least 1 so Conclusion the number of distinct MSTs We’ve now shown that there are at most distinct MSTs.
The claim is proved through induction.