feat(Geometry/Euclidean/Angle/Incenter): unoriented angle bisection#32282
feat(Geometry/Euclidean/Angle/Incenter): unoriented angle bisection#32282jsm28 wants to merge 249 commits intoleanprover-community:masterfrom
Conversation
…Projection_of_le` Add a lemma that `orthogonalProjection s₁ (orthogonalProjection s₂ p) = orthogonalProjection s₁ p` for `s₁ ≤ s₂`. (A similar lemma for projection onto submodules rather than affine subspaces already exists.)
… and bisection lemmas
Add lemmas
```lean
lemma oangle_eq_neg_of_angle_eq_of_sign_eq_neg {w x y z : V}
(h : InnerProductGeometry.angle w x = InnerProductGeometry.angle y z)
(hs : (o.oangle w x).sign = -(o.oangle y z).sign) : o.oangle w x = -o.oangle y z := by
```
and
```lean
lemma angle_eq_iff_oangle_eq_neg_of_sign_eq_neg {w x y z : V} (hw : w ≠ 0) (hx : x ≠ 0)
(hy : y ≠ 0) (hz : z ≠ 0) (hs : (o.oangle w x).sign = -(o.oangle y z).sign) :
InnerProductGeometry.angle w x = InnerProductGeometry.angle y z ↔
o.oangle w x = -o.oangle y z := by
```
and similar affine versions, corresponding to such lemmas that already
exist when the signs are equal rather than negations of each other.
Deduce lemmas relating oriented and unoriented versions of angle
bisection:
```lean
lemma angle_eq_iff_oangle_eq_or_sameRay {x y z : V} (hx : x ≠ 0) (hz : z ≠ 0) :
InnerProductGeometry.angle x y = InnerProductGeometry.angle y z ↔
o.oangle x y = o.oangle y z ∨ SameRay ℝ x z := by
```
and
```lean
lemma angle_eq_iff_oangle_eq_or_wbtw {p₁ p₂ p₃ p₄ : P} (hp₁ : p₁ ≠ p₂) (hp₄ : p₄ ≠ p₂) :
∠ p₁ p₂ p₃ = ∠ p₃ p₂ p₄ ↔ ∡ p₁ p₂ p₃ = ∡ p₃ p₂ p₄ ∨ Wbtw ℝ p₂ p₁ p₄ ∨ Wbtw ℝ p₂ p₄ p₁ := by
```
… equal distance
Add a lemma
```lean
lemma dist_orthogonalProjection_eq_iff_oangle_eq {p p' : P} {s₁ s₂ : AffineSubspace ℝ P}
[s₁.direction.HasOrthogonalProjection] [s₂.direction.HasOrthogonalProjection]
(hp' : p' ∈ s₁ ⊓ s₂)
(hne : haveI : Nonempty s₁ := ⟨p', hp'.1⟩; haveI : Nonempty s₂ := ⟨p', hp'.2⟩;
(orthogonalProjection s₁ p : P) ≠ orthogonalProjection s₂ p)
(hp₁ : haveI : Nonempty s₁ := ⟨p', hp'.1⟩; orthogonalProjection s₁ p ≠ p')
(hp₂ : haveI : Nonempty s₂ := ⟨p', hp'.2⟩; orthogonalProjection s₂ p ≠ p') :
haveI : Nonempty s₁ := ⟨p', hp'.1⟩
haveI : Nonempty s₂ := ⟨p', hp'.2⟩
dist p (orthogonalProjection s₁ p) = dist p (orthogonalProjection s₂ p) ↔
∡ (orthogonalProjection s₁ p : P) p' p = ∡ p p' (orthogonalProjection s₂ p) :=
```
that is an oriented angle analogue of the existing
`dist_orthogonalProjection_eq_iff_angle_eq`. Because the minimal
nondegeneracy conditions required for the two directions of this lemma
are different (whereas the unoriented version doesn't need any
nondegeneracy conditions), those two directions are added as separate
lemmas, each with minimal nondegeneracy conditions, from which the
`iff` version is then deduced.
…ality of twice angles
Add lemmas that two angles less than `π / 2` are equal if and only if
twice those angles are equal, along with a separate lemma
```lean
lemma toReal_neg_eq_neg_toReal_iff {θ : Angle} : (-θ).toReal = -(θ.toReal) ↔ θ ≠ π := by
```
which isn't directly connected but turns out to be useful in the same
application (dealing with angles in right-angled triangles that are
less then `π / 2`, when you have two such triangles that are
oppositely-oriented).
…`π / 2` Add a lemma that an oriented angle in a right-angled triangle is less than `π / 2`, even in degenerate cases (there's already a corresponding lemma for unoriented angles, but that one needs to exclude some degenerate cases because of the different default values for oriented and unoriented angles involving zero vectors). Deduce lemmas that, for such angles involved in right-angled triangles, equality of the angles (i.e. equality mod 2π) follows from equality of twice the angles (i.e. equality of the angles mod π).
Add a lemma
```lean
lemma orthogonalProjection_eq_iff_mem {s : AffineSubspace ℝ P} [Nonempty s]
[s.direction.HasOrthogonalProjection] {p q : P} :
orthogonalProjection s p = q ↔ q ∈ s ∧ p -ᵥ q ∈ s.directionᗮ := by
```
that gives the characteristic property of the orthogonal projection in
a more convenient form to use than the existing
`inter_eq_singleton_orthogonalProjection` (from which it is derived).
…ion_orthogonalProjection_of_le
…isector_dist_oangle
Add instances that the supremum of two affine subspaces, either one nonempty, is nonempty. These are useful when working with orthogonal projections onto such a supremum.
…ion_sup_of_orthogonalProjection_eq
Add a lemma that, if the orthogonal projections of a point onto two subspaces are equal, so is the projection onto their supremum.
Co-authored-by: Eric Wieser <[email protected]>
…sup_of_orthogonalProjection_eq
…ion_orthogonalProjection_of_le
…isector_dist_oangle
…ion_sup_of_orthogonalProjection_eq
…ion_orthogonalProjection_of_le
…isector_dist_oangle
…ion_sup_of_orthogonalProjection_eq
Co-authored-by: Eric Wieser <[email protected]>
…ion_orthogonalProjection_of_le
…isector_dist_oangle
Co-authored-by: Eric Wieser <[email protected]>
Co-authored-by: Eric Wieser <[email protected]>
| lemma angle_incenter_eq_angle_div_two {i₁ i₂ i₃ : Fin 3} (h₁₂ : i₁ ≠ i₂) (h₁₃ : i₁ ≠ i₃) | ||
| (h₂₃ : i₂ ≠ i₃) : | ||
| ∠ (t.points i₂) (t.points i₁) t.incenter = ∠ (t.points i₂) (t.points i₁) (t.points i₃) / 2 := by | ||
| let S : AffineSubspace ℝ P := affineSpan ℝ (Set.range t.points) |
There was a problem hiding this comment.
Random idea and not in the scope of the PR: should we have a "planarize" tactic that handles restricting everything to the subspace?
There was a problem hiding this comment.
Something like that would be reasonable (once we have enough examples to show e.g. what's most convenient in terms of choosing the two-dimensional subspace / showing that all points lie in it / possibly dealing with any special cases where actually the ambient space is less than two-dimensional). Note that we don't yet have a full set of map / restrict lemmas for all simplex definitions, though that might only matter when they're missing for the particular definition you wish to transport to a two-dimensional subspace in a particular case.
Add lemmas giving unoriented angles involving the incenter and excenters of a triangle as expressions involving dividing angles of the triangle by 2, deduced from oriented bisection lemmas.
mapand subtype lemmas #32019mapandrestrictlemmas #32021toRealaddition lemmas #32259mapandrestrictlemmas #32270