Past exam of the mathematics course of the University of Cambridge 2009 ii Paper 2 17F i Solution Created 2026-09-24 Updated 2026-09-24
Past exam of the mathematics course of the University of Cambridge 2009 ii Paper 2 21G Solution Created 2026-09-24 Updated 2026-09-24
Past exam of the mathematics course of the University of Cambridge 2009 ii Paper 3 17F a Solution Created 2026-09-24 Updated 2026-09-24
Past exam of the mathematics course of the University of Cambridge 2009 ii Paper 3 17F b Solution Created 2026-09-24 Updated 2026-09-24
Past exam of the mathematics course of the University of Cambridge 2010 ii Paper 3 3F Solution Created 2026-09-24 Updated 2026-09-24
Past exam of the mathematics course of the University of Cambridge 2012 iii Paper 9 2 ii Solution Created 2026-10-03 Updated 2026-10-07
If some choice produced a graph with largest component of a graph smaller than , the balanced component cut result with packing parameter one would give an empty graph cut whose sides both have at least vertices. For any fixed such vertex set partition, a uniform edge of the complete graph crosses with probabilityTo avoid crossing in row , at least one of its offered edges must be noncrossing. The row's probability is . The rows consist of independent random variables, so at a union bound over at most vertex set partitions givesFor example, the positive integermakes this bound at most . Therefore every choice has a component of order at least , with probability tending to one. The simultaneous giant for fixed random-edge choice estimate includes choices made after seeing the entire array; it does not assume a particular online selection rule.
Past exam of the mathematics course of the University of Cambridge 2012 iii Paper 9 5 i Solution Created 2026-10-03 Updated 2026-10-07
Use the standard Achlioptas process with two independent uniform candidate edges per step and a rule choosing one. The component-band vertex count counts vertices, not the number of graph components. Integer time rounding changes the arguments below by steps.
First, at the eventual graph component of order at least is assembled from graph components already present at . Contract those initial graph components. During steps, at most new edges are added, so the eventual connected contracted graph contains at most initial graph components. Initial graph components of order less than can therefore contribute at mostvertices, for sufficiently large and fixed , since . For their contribution is zero. Consequently .
We now use forced merging of large components. If at a time the graph components of order at least contain at least vertices, then after at most steps there is a graph component of order at least , except on an event of probability exponentially small in . One explicit choice is . The proof is a union bound over set partitions of the at most initial large graph components: if all final graph components were small, a balanced component cut would split their initial vertices into sets of size at least , and both candidate edges cross that fixed cut with probability at least in every step. The rule cannot avoid such a forced crossing. The failure bound is , and a union bound makes the estimate simultaneous over all starting times at most , for each fixed .