AFTD Auto-Formalizing Theoretical Domains

root_exposed_fort_iff_new_white_leaf_breaks_forcing

theoremverified

If the complement of W is a zero forcing set of G, then W contains a root-exposed fort exactly when the same originally blue vertices fail to force the graph obtained by adjoining an initially white leaf at r.

Statement

theorem root_exposed_fort_iff_new_white_leaf_breaks_forcing {V : Type*} [Fintype V] [DecidableEq V] (G : SimpleGraph V) (r : V) (W : Finset V) (hW : is_zero_forcing_set G ((Finset.univ \ W : Finset V) : Set V)) : zero_forcing_root_exposed_fort G r W ↔ ¬ is_zero_forcing_set (zero_forcing_attach_leaf G r) (Option.some '' (((Finset.univ \ W : Finset V) : Set V)))
Source
The operational meaning of the rooted boundary state in OP-1399; a related leaf inequality is in arXiv:2605.10836, Section 3, Lemma 4.
Verified
13 Nov 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 4
  • zero_forcing_root_exposed_fortA white set W contains a root-exposed fort if it contains a set F not containing r such that r has a neighbor in F and every vertex other…
  • zero_forcing_attach_leafThe graph obtained from G by adjoining one new vertex adjacent only to the root r.
  • zfs_compl_iff_no_stuckFor any finite graph G and finite T, the complement V\T is a zero forcing set precisely when T contains no nonempty "stuck" subset W, i.e.…
  • is_zero_forcing_setA set of vertices S is a zero forcing set of G if its zero forcing closure contains all vertices of G.

The proof, explained

By the fort characterization of zero forcing (`zfs_compl_iff_no_stuck`), the complement of W forces G if and only if W contains no fort, and the extended white set W_ext = W ∪ {none} fails to force the leaf-augmented graph if and only if W_ext contains a fort W₁. In the forward direction, given a root-exposed fort F ⊆ W, adjoining the new leaf gives a fort W₁ = F ∪ {none} in the extended graph, since any neighbor of none must be the root r (which has another neighbor in F), and r has none as a second neighbor in W₁. In the reverse direction, given a fort W₁ ⊆ W_ext, the hypothesis that V \ W forces G prevents W₁ from avoiding none or containing some r, which forces none ∈ W₁ and r ∉ to_v(W₁); the projection to_v(W₁) is then shown to be a root-exposed fort in G.

  1. An embedded vertex some v belongs to the extended white set W_ext if and only if v ∈ W. automation

    By unfolding the definition of W_ext = {none} ∪ {some w | w ∈ W}, since some v ≠ none and some is injective, some v ∈ W_ext simplifies directly to v ∈ W.

    lines 7–10

    lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some
        {V : Type*} [DecidableEq V] {v : V} {W : Finset V} :
        some v ∈ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W ↔ v ∈ W := by
      simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]
  2. A vertex v belongs to the projection to_v(W₁) if and only if some v ∈ W₁. automation

    The projection to_v(W₁) is defined by filtering the vertices v of V such that some v ∈ W₁; simplifying with this definition yields the claim directly.

    lines 16–19

    lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v
        {V : Type*} [Fintype V] [DecidableEq V] {W1 : Finset (Option V)} {v : V} :
        v ∈ root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 ↔ some v ∈ W1 := by
      simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v]
  3. If W₁ is a subset of the extended white set W_ext, then its projection to_v(W₁) is a subset of W.

    For any v ∈ to_v(W₁), we have some v ∈ W₁ by root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v. Since W₁ ⊆ W_ext, this gives some v ∈ W_ext, which by root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some implies v ∈ W.

    lines 21–28

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some

    lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub
        {V : Type*} [Fintype V] [DecidableEq V] {W1 : Finset (Option V)} {W : Finset V}
        (h : W1 ⊆ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W) :
        root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 ⊆ W := by
      intro v hv
      rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hv
      have := h hv
      rwa [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some] at this
  4. The originally blue vertices in the leaf-augmented graph equal the complement of the extended white set W_ext. automation

    By extensionality on Option V, the leaf none belongs to neither set, and for any vertex x ∈ V, some x belongs to some '' (V \ W) if and only if x ∉ W, which is equivalent to some x ∉ W_ext.

    lines 30–36

    lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq
        {V : Type*} [Fintype V] [DecidableEq V] (W : Finset V) :
        (Option.some '' (((Finset.univ \ W : Finset V) : Set V))) =
        ((Finset.univ \ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W : Finset (Option V)) : Set (Option V)) := by
      ext (x | x)
      · simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]
      · simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]
  5. To show that the leaf-augmented graph fails to be forced, it suffices to construct a non-empty fort contained in W_ext from the assumed root-exposed fort F ⊆ W.

    We split the theorem into two implications. In the forward direction, assuming W contains a root-exposed fort F, we rewrite the failure of zero forcing using zfs_compl_iff_no_stuck and root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq: it is equivalent to the existence of a non-empty fort in the extended graph contained in W_ext.

    lines 38–44

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck

    theorem root_exposed_fort_iff_new_white_leaf_breaks_forcing {V : Type*} [Fintype V] [DecidableEq V] (G : SimpleGraph V) (r : V) (W : Finset V) (hW : is_zero_forcing_set G ((Finset.univ \ W : Finset V) : Set V)) : zero_forcing_root_exposed_fort G r W ↔ ¬ is_zero_forcing_set (zero_forcing_attach_leaf G r) (Option.some '' (((Finset.univ \ W : Finset V) : Set V))) := by
      constructor
      · intro hF
        rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck]
        intro h
        rcases hF with ⟨F, hFW, hrF, ⟨w0, hw0F, hrw0⟩, hfort⟩
        apply h
  6. The candidate set W₁ = {none} ∪ {some f | f ∈ F} is non-empty and contained in W_ext.

    The set W₁ is non-empty by Finset.insert_nonempty because it contains none. Since F ⊆ W, any element of W₁ is either none or some y with y ∈ W, so W₁ ⊆ W_ext.

    lines 45–52

    uses Finset.insert_nonempty

        refine ⟨insert none (F.map ⟨Option.some, Option.some_injective V⟩), ?_, ?_, ?_⟩
        · exact Finset.insert_nonempty none _
        · intro x hx
          simp only [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf,
            mem_insert, mem_map, Function.Embedding.coeFn_mk] at hx ⊢
          rcases hx with rfl | ⟨y, hy, rfl⟩
          · left; rfl
          · right; exact ⟨y, hFW hy, rfl⟩
  7. Any vertex u outside W₁ cannot be the leaf none, and if u is adjacent to the leaf w = none, then u must be the root r, which has an alternate neighbor some w₀ in W₁.

    Let u ∉ W₁ be adjacent to w ∈ W₁. Since none ∈ W₁, u cannot be none, so u = some u' for u' ∈ V. If w = none, adjacency in the leaf-augmented graph forces u' = r. Since F is a root-exposed fort, r has a neighbor w₀ ∈ F, so some w₀ ∈ W₁ provides a second neighbor of some r in W₁ distinct from none.

    lines 53–66

        · intro u hu w hw hadj
          rcases u with _ | u
          · exfalso
            apply hu
            simp only [mem_insert, true_or]
          · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk] at hw
            rcases hw with rfl | ⟨w', hw', rfl⟩
            · have hu_eq_r : u = r := hadj
              rw [hu_eq_r]
              refine ⟨some w0, ?_, ?_, ?_⟩
              · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]
                right; exact ⟨w0, hw0F, rfl⟩
              · exact hrw0
              · intro h; cases h
  8. If w = some w' with w' ∈ F, the vertex u has an alternate neighbor in W₁: none if u' = r, or some x' from the root-exposed fort condition if u' ≠ r, completing the forward direction.

    If w = some w' with w' ∈ F, then u' ∉ F and u' is adjacent to w' in G. If u' = r, the leaf none ∈ W₁ serves as an alternate neighbor of some r distinct from w. If u' ≠ r, the root-exposed fort property of F provides an alternate neighbor x' ∈ F of u' with x' ≠ w', so some x' ∈ W₁ is adjacent to some u' and distinct from w. Thus W₁ is a fort in the extended graph.

    lines 67–86

            · have hadj_G : G.Adj u w' := hadj
              have huF : u ∉ F := by
                intro huF
                apply hu
                simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]
                right; exact ⟨u, huF, rfl⟩
              by_cases hur : u = r
              · rw [hur]
                refine ⟨none, ?_, ?_, ?_⟩
                · simp only [mem_insert, true_or]
                · show r = r; rfl
                · intro h; cases h
              · rcases hfort u hur huF w' hw' hadj_G with ⟨x', hx'F, hadjx', hne⟩
                refine ⟨some x', ?_, ?_, ?_⟩
                · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]
                  right; exact ⟨x', hx'F, rfl⟩
                · exact hadjx'
                · intro h
                  injection h with h'
                  exact hne h'
  9. In the reverse direction, assuming the leaf-augmented graph is not forced, there exists a non-empty fort W₁ ⊆ W_ext, while W contains no forts in G.

    Applying zfs_compl_iff_no_stuck and root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq to the negation of zero forcing yields a non-empty fort W₁ ⊆ W_ext in the leaf-augmented graph. Applying zfs_compl_iff_no_stuck to the hypothesis hW shows that G contains no non-empty fort contained in W.

    lines 87–96

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck

      · intro h_not
        rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck] at h_not
        have h_ex : ∃ W1 : Finset (Option V), W1.Nonempty ∧
            W1 ⊆ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W ∧
            ∀ u ∉ W1, ∀ w ∈ W1, (zero_forcing_attach_leaf G r).Adj u w →
              ∃ x ∈ W1, (zero_forcing_attach_leaf G r).Adj u x ∧ x ≠ w := by
          by_contra hc
          exact h_not hc
        rcases h_ex with ⟨W1, hW1_ne, hW1_sub, hW1_fort⟩
        rw [zfs_compl_iff_no_stuck] at hW
  10. The leaf none must belong to W₁.

    If none ∉ W₁, then every element of W₁ is of the form some v, so the projection to_v(W₁) is non-empty and contained in W by root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub. For any u ∉ to_v(W₁) adjacent to w ∈ to_v(W₁) in G, the fort property of W₁ provides an alternate neighbor in W₁; since none ∉ W₁, this alternate neighbor must be some x_v, meaning x_v ∈ to_v(W₁) is an alternate neighbor in G. Thus to_v(W₁) would be a fort in G inside W, contradicting hW.

    lines 97–119

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v

        have h_none_in : none ∈ W1 := by
          by_contra h_none_notin
          apply hW
          refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1, ?_,
            root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, ?_⟩
          · rcases hW1_ne with ⟨x, hx⟩
            cases x with
            | none => contradiction
            | some v =>
              exact ⟨v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx⟩
          · intro u hu w hw hadj
            rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw
            have hu_not_in : some u ∉ W1 := fun h => hu
              (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)
            have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj
            rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩
            cases x with
            | none => contradiction
            | some x_v =>
              refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩
              intro heq
              subst heq
              exact hxne rfl
  11. The root r does not belong to to_v(W₁).

    If r ∈ to_v(W₁), then to_v(W₁) is non-empty and contained in W. For any u ∉ to_v(W₁) adjacent to w ∈ to_v(W₁), the fort property of W₁ provides an alternate neighbor x ∈ W₁. If x were none, adjacency would force u = r, impossible since r ∈ to_v(W₁) and u ∉ to_v(W₁). Thus x = some x_v, giving an alternate neighbor in to_v(W₁). This would make to_v(W₁) a fort in G inside W, again contradicting hW.

    lines 120–140

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v

        have hr_notin : r ∉ root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 := by
          intro hr_in
          apply hW
          refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1,
            ⟨r, hr_in⟩, root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, ?_⟩
          intro u hu w hw hadj
          rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw
          have hu_not_in : some u ∉ W1 := fun h => hu
            (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)
          have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj
          rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩
          cases x with
          | none =>
            have hu_eq_r : u = r := hadjx
            subst hu_eq_r
            exact (hu hr_in).elim
          | some x_v =>
            refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩
            intro heq
            subst heq
            exact hxne rfl
  12. The set to_v(W₁) is contained in W, does not contain r, and contains a neighbor of r.

    We already know to_v(W₁) ⊆ W and r ∉ to_v(W₁). Because r ∉ to_v(W₁), some r ∉ W₁, while none ∈ W₁ and some r is adjacent to none. By the fort property of W₁, some r has an alternate neighbor x ∈ W₁ distinct from none. Hence x = some w with w ∈ to_v(W₁), showing that r has a neighbor w in to_v(W₁).

    lines 141–150

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v

        refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1,
          root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, hr_notin, ?_, ?_⟩
        · have hr_notin' : some r ∉ W1 := fun h => hr_notin
            (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)
          have hadj_leaf : (zero_forcing_attach_leaf G r).Adj (some r) none := rfl
          rcases hW1_fort (some r) hr_notin' none h_none_in hadj_leaf with ⟨x, hx, hadjx, hxne⟩
          cases x with
          | none => exact (hxne rfl).elim
          | some w =>
            refine ⟨w, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx⟩
  13. For every u ≠ r outside to_v(W₁), any neighbor in to_v(W₁) has a companion neighbor in to_v(W₁), proving that to_v(W₁) is a root-exposed fort.

    Let u ≠ r with u ∉ to_v(W₁), and let w ∈ to_v(W₁) be adjacent to u in G. Then some u ∉ W₁ is adjacent to some w ∈ W₁. The fort property of W₁ yields an alternate neighbor x ∈ W₁ distinct from some w. Since u ≠ r, x cannot be none, so x = some x_v for some x_v ∈ to_v(W₁) adjacent to u and distinct from w. This establishes that to_v(W₁) is a root-exposed fort in W.

    lines 151–165

    uses root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v

        · intro u hur hu w hw hadj
          have hu_not_in : some u ∉ W1 := fun h => hu
            (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)
          rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw
          have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj
          rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩
          cases x with
          | none =>
            have hu_eq_r : u = r := hadjx
            exact (hur hu_eq_r).elim
          | some x_v =>
            refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩
            intro heq
            subst heq
            exact hxne rfl
Finset.insert_nonempty
Inserting an element into a finite set produces a non-empty set.
root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq
The set of originally blue vertices in the graph with the attached leaf is equal to the complement of the extended white set W_ext.
root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some
A vertex v from the original graph, embedded as some v, belongs to the extended white set W_ext if and only if v ∈ W.
root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v
A vertex v belongs to the projection to_v(W₁) if and only if some v belongs to W₁.
root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub
If a vertex subset W₁ of the extended graph is contained in the extended white set W_ext, then its projection to_v(W₁) is contained in W.
zfs_compl_iff_no_stuck
The complement of a vertex set T is a zero forcing set in G if and only if T contains no non-empty fort (stuck set).

Written by a model (gemini-3.8-flash) from the Lean proof and the goal states Lean reported at each step. Lean verified the proof; this explanation is a reading aid and is not itself machine-checked.

Lean source view module on GitHub

open Finsetdef root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf    {V : Type*} [DecidableEq V] (W : Finset V) : Finset (Option V) :=  insert none (W.map ⟨Option.some, Option.some_injective V⟩)lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some    {V : Type*} [DecidableEq V] {v : V} {W : Finset V} :    some v ∈ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W ↔ v ∈ W := by  simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]def root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v    {V : Type*} [Fintype V] [DecidableEq V] (W1 : Finset (Option V)) : Finset V :=  Finset.univ.filter (fun v => some v ∈ W1)lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v    {V : Type*} [Fintype V] [DecidableEq V] {W1 : Finset (Option V)} {v : V} :    v ∈ root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 ↔ some v ∈ W1 := by  simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v]lemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub    {V : Type*} [Fintype V] [DecidableEq V] {W1 : Finset (Option V)} {W : Finset V}    (h : W1 ⊆ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W) :    root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 ⊆ W := by  intro v hv  rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hv  have := h hv  rwa [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_t_leaf_some] at thislemma root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq    {V : Type*} [Fintype V] [DecidableEq V] (W : Finset V) :    (Option.some '' (((Finset.univ \ W : Finset V) : Set V))) =    ((Finset.univ \ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W : Finset (Option V)) : Set (Option V)) := by  ext (x | x)  · simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]  · simp [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf]theorem root_exposed_fort_iff_new_white_leaf_breaks_forcing {V : Type*} [Fintype V] [DecidableEq V] (G : SimpleGraph V) (r : V) (W : Finset V) (hW : is_zero_forcing_set G ((Finset.univ \ W : Finset V) : Set V)) : zero_forcing_root_exposed_fort G r W ↔ ¬ is_zero_forcing_set (zero_forcing_attach_leaf G r) (Option.some '' (((Finset.univ \ W : Finset V) : Set V))) := by  constructor  · intro hF    rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck]    intro h    rcases hF with ⟨F, hFW, hrF, ⟨w0, hw0F, hrw0⟩, hfort⟩    apply h    refine ⟨insert none (F.map ⟨Option.some, Option.some_injective V⟩), ?_, ?_, ?_⟩    · exact Finset.insert_nonempty none _    · intro x hx      simp only [root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf,        mem_insert, mem_map, Function.Embedding.coeFn_mk] at hx ⊢      rcases hx with rfl | ⟨y, hy, rfl⟩      · left; rfl      · right; exact ⟨y, hFW hy, rfl⟩    · intro u hu w hw hadj      rcases u with _ | u      · exfalso        apply hu        simp only [mem_insert, true_or]      · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk] at hw        rcases hw with rfl | ⟨w', hw', rfl⟩        · have hu_eq_r : u = r := hadj          rw [hu_eq_r]          refine ⟨some w0, ?_, ?_, ?_⟩          · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]            right; exact ⟨w0, hw0F, rfl⟩          · exact hrw0          · intro h; cases h        · have hadj_G : G.Adj u w' := hadj          have huF : u ∉ F := by            intro huF            apply hu            simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]            right; exact ⟨u, huF, rfl⟩          by_cases hur : u = r          · rw [hur]            refine ⟨none, ?_, ?_, ?_⟩            · simp only [mem_insert, true_or]            · show r = r; rfl            · intro h; cases h          · rcases hfort u hur huF w' hw' hadj_G with ⟨x', hx'F, hadjx', hne⟩            refine ⟨some x', ?_, ?_, ?_⟩            · simp only [mem_insert, mem_map, Function.Embedding.coeFn_mk]              right; exact ⟨x', hx'F, rfl⟩            · exact hadjx'            · intro h              injection h with h'              exact hne h'  · intro h_not    rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_compl_t_leaf_eq, zfs_compl_iff_no_stuck] at h_not    have h_ex : ∃ W1 : Finset (Option V), W1.Nonempty ∧        W1 ⊆ root_exposed_fort_iff_new_white_leaf_breaks_forcing_t_leaf W ∧        ∀ u ∉ W1, ∀ w ∈ W1, (zero_forcing_attach_leaf G r).Adj u w →          ∃ x ∈ W1, (zero_forcing_attach_leaf G r).Adj u x ∧ x ≠ w := by      by_contra hc      exact h_not hc    rcases h_ex with ⟨W1, hW1_ne, hW1_sub, hW1_fort⟩    rw [zfs_compl_iff_no_stuck] at hW    have h_none_in : none ∈ W1 := by      by_contra h_none_notin      apply hW      refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1, ?_,        root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, ?_⟩      · rcases hW1_ne with ⟨x, hx⟩        cases x with        | none => contradiction        | some v =>          exact ⟨v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx⟩      · intro u hu w hw hadj        rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw        have hu_not_in : some u ∉ W1 := fun h => hu          (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)        have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj        rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩        cases x with        | none => contradiction        | some x_v =>          refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩          intro heq          subst heq          exact hxne rfl    have hr_notin : r ∉ root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1 := by      intro hr_in      apply hW      refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1,        ⟨r, hr_in⟩, root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, ?_⟩      intro u hu w hw hadj      rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw      have hu_not_in : some u ∉ W1 := fun h => hu        (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)      have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj      rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩      cases x with      | none =>        have hu_eq_r : u = r := hadjx        subst hu_eq_r        exact (hu hr_in).elim      | some x_v =>        refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩        intro heq        subst heq        exact hxne rfl    refine ⟨root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v W1,      root_exposed_fort_iff_new_white_leaf_breaks_forcing_to_v_sub hW1_sub, hr_notin, ?_, ?_⟩    · have hr_notin' : some r ∉ W1 := fun h => hr_notin        (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)      have hadj_leaf : (zero_forcing_attach_leaf G r).Adj (some r) none := rfl      rcases hW1_fort (some r) hr_notin' none h_none_in hadj_leaf with ⟨x, hx, hadjx, hxne⟩      cases x with      | none => exact (hxne rfl).elim      | some w =>        refine ⟨w, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx⟩    · intro u hur hu w hw hadj      have hu_not_in : some u ∉ W1 := fun h => hu        (root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr h)      rw [root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v] at hw      have hadj' : (zero_forcing_attach_leaf G r).Adj (some u) (some w) := hadj      rcases hW1_fort (some u) hu_not_in (some w) hw hadj' with ⟨x, hx, hadjx, hxne⟩      cases x with      | none =>        have hu_eq_r : u = r := hadjx        exact (hur hu_eq_r).elim      | some x_v =>        refine ⟨x_v, root_exposed_fort_iff_new_white_leaf_breaks_forcing_mem_to_v.mpr hx, hadjx, ?_⟩        intro heq        subst heq        exact hxne rfl
See it in the graphDiscuss this resultChallenge it