13.07.2015 Views

The Nash-Kuiper C -isometric embedding theorem in exercises

The Nash-Kuiper C -isometric embedding theorem in exercises

The Nash-Kuiper C -isometric embedding theorem in exercises

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>The</strong> <strong>Nash</strong>-<strong>Kuiper</strong> C 1 -<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong> <strong>theorem</strong><strong>in</strong> <strong>exercises</strong>Decomposition <strong>in</strong>to primitive metricsWe use the term Riemannian semi-metric for a field of non-negative quadraticforms on the tangent bundle of a manifold.A Riemannian semi-metric g on R n is called primitive if g = α(x)(dl) 2 ,where l = l(x) is a l<strong>in</strong>ear function on R n and α is a non-negative functionwith compact support.A Riemannian semi-metric g on a manifold V is called primitive if thereexists a local parametrization u : R n → U ⊂ V such that supp g ⊂ U andu ∗ g is a primitive semi-metric on R n .1. Prove that any Riemannian semi-metric g on a compact manifold V canbe decomposed <strong>in</strong>to a f<strong>in</strong>ite sum of primitive semi-metrics.Smoothness of a limitLet g be a Riemannian metric on a manifold V and h the Euclidean metricon R q . Given two maps f, ˜f : V → R q we denote by d g (f, ˜f) the functiond g (f, ˜f)TV \ V → R , v → d g (f, ˜f)(v) = || f ∗v − ˜f ∗ v|| h|| v|| g,where h is the standard metric on R q .2. Let f i : V → R q Cbe a sequence of (smooth) maps. Prove that if f 0i →f andd g (f i , f i+1 ) < c i , with ∑ c i < ∞ , then f is a C 1 C-smooth map and f i →f 1.i<strong>The</strong> function r(˜g, g)Given a positive metric g and non-negative ˜g we denote by r(˜g, g) the functionTV \ V → R , v → r(˜g, g)(v) = || v|| eg|| v|| g.3. Show that that1


(r1) r(˜g, g 1 ) ≤ r(˜g, g 2 ) if g 1 ≥ g 2and(r2) r(˜g, g) ≤ r(˜g + g 1 , g + g 1 ) if r(˜g, g) ≤ 1 .One-dimensional approximationAn <strong>embedd<strong>in</strong>g</strong> (V, g) → (R q , h) is called short (resp. strictly short) if theform g − f ∗ h ≤ h is non-negative, i.e. semi-metric (resp. positive def<strong>in</strong>ite).4. Let g be any Riemannian metric on the <strong>in</strong>terval I. Prove that if q > 1then for any ε > 0 any short <strong>embedd<strong>in</strong>g</strong> f : (I, g) → (R q , h) can be C 0 -approximated by ε-<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong>s. Moreover, for any ρ > 0 the approximat<strong>in</strong>gmap ˜f can be chosen <strong>in</strong> such a way thatwhere ∆ = g − f ∗ h.Add<strong>in</strong>g a primitive metricd g (f, ˜f) < r(∆, g) + ρProve the follow<strong>in</strong>g generalization of Exercise 4:5. Suppose that dim V = n < q. Let f : (V, g) → (R q , h) be a short <strong>embedd<strong>in</strong>g</strong>such that ∆ = g − f ∗ h is a primitive semi-metric on V . <strong>The</strong>n for any ε the<strong>embedd<strong>in</strong>g</strong> f can be C 0 -approximated by ε-<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong>s. Moreover,for any ρ > 0 the approximat<strong>in</strong>g map ˜f can be chosen to satisfy the <strong>in</strong>equalityApproximation <strong>The</strong>oremd g (f, ˜f) < r(∆, g) + ρ .An <strong>embedd<strong>in</strong>g</strong> f : (V, g) → (R q , h) is called ε-<strong>isometric</strong> if(1 − ε)g < f ∗ h < (1 + ε)g .Prove the follow<strong>in</strong>g Approximation <strong>theorem</strong>:6. Let n < q. For any ε > 0, any short <strong>embedd<strong>in</strong>g</strong> f : (V n , g) → (R q , h)can be C 0 -approximated by ε-<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong>s. Moreover, given a fixeddecomposition of a Riemannian semi-metric∆ = g − f ∗ h2


<strong>in</strong>to a sum of N primitive metrics then for any constant ρ > 0 one can choosean approximat<strong>in</strong>g <strong>embedd<strong>in</strong>g</strong> ˜f which satisfies the <strong>in</strong>equalityd g (f, ˜f) < N r(∆, g) + ρ .<strong>Nash</strong>-<strong>Kuiper</strong>’s C 1 -<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong> <strong>theorem</strong>Deduce from Exercises 6 and 2 the <strong>Nash</strong>-<strong>Kuiper</strong> C 1 -<strong>isometric</strong> <strong>embedd<strong>in</strong>g</strong><strong>theorem</strong>:7. If n < q then any strictly short immersionf : (V n , g) → (R q , h) ,where h is the standard metric on R q , can be C 0 -approximated by <strong>isometric</strong>C 1 -smooth immersions. Moreover, if the <strong>in</strong>itial immersion f is an <strong>embedd<strong>in</strong>g</strong>then f can be approximated by <strong>isometric</strong> C 1 -<strong>embedd<strong>in</strong>g</strong>s.Bound for the regularity <strong>in</strong> the <strong>Nash</strong>-<strong>Kuiper</strong> <strong>theorem</strong>8. Show that the h-pr<strong>in</strong>ciple type results fail for C 2 -<strong>isometric</strong> immersions.3

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!