How to linearize your cluster

Thank you, now I understand. All of this started in my head while trying to understand the proof of the following.

I tried to expand the proof of it, using the additional material you provided. From the comment above we can deduce the following

Lemma.

  1. pot is the (i.e. unique) highest amongst all the sets including inc and excluding exc.
  2. t\in pot if and only if feerate(t)> feerate(pot).

Proposition. Let B an element that maximize the feerate in the following collection.\{U \text{ topological} \mid inc \subseteq U \subseteq exc^c \}. If C \subseteq pot is topological, then C\subseteq B.

Proof. By contraddiction, suppose that there is B like in the hypothesis and C topological, C\subseteq pot and C \not \subseteq B. Since C is topological, possibly replacing t with some of its ancestors, it is no loss to assume anc(t) \subseteq B. Since t\in pot, by 2. of Lemma above, feerate(t) > feerate(pot). By characterization of pot in 1. of Lemma above, we have that feerate(pot) \ge feerate(B). So, feerate(t)> feerate(B). Consider B' = \{t\} \cup B. Clearly, feerate(B')> feerate(B); moreover, since anc(t) \subseteq (B), B' is topological. This is against the definition of B.

Let me know what do you think.