Riemannian manifolds #
A Riemannian manifold M
is a real manifold such that its tangent spaces are endowed with an
inner product, depending smoothly on the point, and such that M
has an emetric space
structure for which the distance is the infimum of lengths of paths.
We register a Prop-valued typeclass IsRiemannianManifold I M
recording this fact, building on top
of [EMetricSpace M] [RiemannianBundle (fun (x : M) โฆ TangentSpace I x)]
.
We show that an inner product vector space, with the associated canonical Riemannian metric,
satisfies the predicate IsRiemannianManifold ๐(โ, E) E
.
In a general manifold with a Riemannian metric, we define the associated extended distance in the
manifold, and show that it defines the same topology as the pre-existing one. Therefore, one
may endow the manifold with an emetric space structure, see EmetricSpace.ofRiemannianMetric
.
By definition, it then satisfies the predicate IsRiemannianManifold I M
.
The following code block is the standard way to say "Let M
be a C^โ
Riemannian manifold".
variable
{E : Type*} [NormedAddCommGroup E] [NormedSpace โ E]
{H : Type*} [TopologicalSpace H] {I : ModelWithCorners โ E H}
{M : Type*} [EMetricSpace M] [ChartedSpace H M] [IsManifold I โ M]
[RiemannianBundle (fun (x : M) โฆ TangentSpace I x)]
[IsContMDiffRiemannianBundle I โ E (fun (x : M) โฆ TangentSpace I x)]
[IsRiemannianManifold I M]
To register a C^n
manifold for a general n
, one should replace [IsManifold โ I M]
with
[IsManifold I n M] [IsManifold I 1 M]
, where the second one is needed to ensure that the
tangent bundle is well behaved (not necessary when n
is concrete like 2 or 3 as there are
automatic instances for these cases). One can require whatever regularity one wants in the
IsContMDiffRiemannianBundle
instance above, for example
[IsContMDiffRiemannianBundle I n E (fun (x : M) โฆ TangentSpace I x)]
, and one should also add
[IsContinuousRiemannianBundle E (fun (x : M) โฆ TangentSpace I x)]
(as above, Lean can not infer
the latter from the former as it can not guess n
).
Consider a manifold in which the tangent spaces are already endowed with an inner product, and
the space is already endowed with an extended distance. We say that this is a Riemannian manifold
if the distance is given by the infimum of the lengths of C^1
paths, measured using the norm in
the tangent spaces.
This is a Prop
valued typeclass, on top of existing data.
If you need to construct a distance using a Riemannian structure,
see EmetricSpace.ofRiemannianMetric
.
Instances
Riemannian structure on an inner product vector space #
We endow an inner product vector space with the canonical Riemannian metric, given by the
inner product of the vector space in each of the tangent spaces, and we show that this construction
satisfies the IsRiemannianManifold ๐(โ, E) E
predicate, i.e., the extended distance between
two points is the infimum of the length of paths between these points.
The standard riemannian metric on a vector space with an inner product, given by this inner product on each tangent space.
Equations
Instances For
An inner product vector space is a Riemannian manifold, i.e., the distance between two points is the infimum of the lengths of paths between these points.
Constructing a distance from a Riemannian structure #
Let M
be a real manifold with a Riemannian structure. We construct the associated distance and
show that the associated topology coincides with the pre-existing topology. Therefore, one may
endow M
with an emetric space structure, called EmetricSpace.ofRiemannianMetric
.
Moreover, we show that in this case the resulting emetric space satisfies the predicate
IsRiemannianManifold I M
.
Showing that the distance topology coincides with the pre-existing topology is not trivial. The
two inclusions are proved respectively in eventually_riemmanianEDist_lt
and
setOf_riemmanianEDist_lt_subset_nhds
.
For the first one, we have to show that points which are close for the topology are at small distance. For this, we use the path between the two points which is the pullback of the segment in the extended chart, and argue that it is short because the images are close in the extended chart.
For the second one, we have to show that any neighborhood of x
contains all the points y
with riemannianEDist x y < c
for some c > 0
. For this, we argue that a short path from x
to y
remains short in the extended chart, and therefore it doesn't have the time to exit
the image of the neighborhood in the extended chart.
Register on the tangent space to a normed vector space the same NormedAddCommGroup
structure
as in the vector space.
Should not be a global instance, as it does not coincide definitionally with the Riemannian structure for inner product spaces, but can be activated locally.
Instances For
Register on the tangent space to a normed vector space the same NormedSpace
structure
as in the vector space.
Should not be a global instance, as it does not coincide definitionally with the Riemannian structure for inner product spaces, but can be activated locally.
Equations
Instances For
Around any point x
, the Riemannian distance between two points is controlled by the distance
in the extended chart. In other words, the extended chart is locally Lipschitz.
If points are close for the topology, then their Riemannian distance is small.
Any neighborhood of x
contains all the points which are close enough to x
for the
Riemannian distance, โโฅ0
version.
Any neighborhood of x
contains all the points which are close enough to x
for the
Riemannian distance, โโฅ0โ
version.
The pseudoemetric space structure associated to a Riemannian metric on a manifold. Designed so that the topology is defeq to the original one.
This should only be used when constructing data in specific situtations. To develop the theory,
one should rather assume that there is an already existing emetric space structure, which satisfies
additionally the predicate IsRiemannianManifold I M
.
Equations
- PseudoEmetricSpace.ofRiemannianMetric I M = PseudoEmetricSpace.ofEdistOfTopology (Manifold.riemannianEDist I) โฏ โฏ โฏ โฏ
Instances For
Given a manifold with a Riemannian metric, consider the associated Riemannian distance. Then
by definition the distance is the infimum of the length of paths between the points, i.e., the
manifold satsifies the IsRiemannianManifold I M
predicate.
The emetric space structure associated to a Riemannian metric on a manifold. Designed so that the topology is defeq to the original one.
This should only be used when constructing data in specific situations. To develop the theory,
one should rather assume that there is an already existing emetric space structure, which satisfies
additionally the predicate IsRiemannianManifold I M
.