Eisenbud–Levine–Khimshiashvili signature formula: Difference between revisions
m →Example: tidy center tags |
|||
Line 41: | Line 41: | ||
The first step for finding the non-degenerate, bilinear form β is to calculate the multiplication table of ''B<sub>X</sub>''; reducing each entry modulo ''I<sub>X</sub>''. Whence |
The first step for finding the non-degenerate, bilinear form β is to calculate the multiplication table of ''B<sub>X</sub>''; reducing each entry modulo ''I<sub>X</sub>''. Whence |
||
{| class="wikitable" style="text-align:center;" |
{| class="wikitable" style="text-align:center;margin:1em auto;" |
||
|- |
|- |
||
!∗ |
!∗ |
Revision as of 15:39, 16 March 2022
In mathematics, and especially differential topology and singularity theory, the Eisenbud–Levine–Khimshiashvili signature formula gives a way of computing the Poincaré–Hopf index of a real, analytic vector field at an algebraically isolated singularity.[1][2] It is named after David Eisenbud, Harold I. Levine, and George Khimshiashvili. Intuitively, the index of a vector field near a zero is the number of times the vector field wraps around the sphere. Because analytic vector fields have a rich algebraic structure, the techniques of commutative algebra can be brought to bear to compute their index. The signature formula expresses the index of an analytic vector field in terms of the signature of a certain quadratic form.
Nomenclature
Consider the n-dimensional space Rn. Assume that Rn has some fixed coordinate system, and write x for a point in Rn, where x = (x1, …, xn).
Let X be a vector field on Rn. For 1 ≤ k ≤ n there exist functions ƒk : Rn → R such that one may express X as
To say that X is an analytic vector field means that each of the functions ƒk : Rn → R is an analytic function. One says that X is singular at a point p in Rn (or that p is a singular point of X) if X(p) = 0, i.e. X vanishes at p. In terms of the functions ƒk : Rn → R it means that ƒk(p) = 0 for all 1 ≤ k ≤ n. A singular point p of X is called isolated (or that p is an isolated singularity of X) if X(p) = 0 and there exists an open neighbourhood U ⊆ Rn, containing p, such that X(q) ≠ 0 for all q in U, different from p. An isolated singularity of X is called algebraically isolated if, when considered over the complex domain, it remains isolated.[3][4]
Since the Poincaré–Hopf index at a point is a purely local invariant (cf. Poincaré–Hopf theorem), one may restrict one's study to that of germs. Assume that each of the ƒk from above are function germs, i.e. ƒk : (Rn,0) → (R,0). In turn, one may call X a vector field germ.
Construction
Let An,0 denote the ring of analytic function germs (Rn,0) → (R,0). Assume that X is a vector field germ of the form
with an algebraically isolated singularity at 0. Where, as mentioned above, each of the ƒk are function germs (Rn,0) → (R,0). Denote by IX the ideal generated by the ƒk, i.e. IX = (ƒ1, …, ƒn). Then one considers the local algebra, BX, given by the quotient
The Eisenbud–Levine–Khimshiashvili signature formula states that the index of the vector field X at 0 is given by the signature of a certain non-degenerate bilinear form (to be defined below) on the local algebra BX.[2][4][5]
The dimension of is finite if and only if the complexification of X has an isolated singularity at 0 in Cn; i.e. X has an algebraically isolated singularity at 0 in Rn.[2] In this case, BX will be a finite-dimensional, real algebra.
Definition of the bilinear form
Using the analytic components of X, one defines another analytic germ F : (Rn,0) → (Rn,0) given by
for all x ∈ Rn. Let JF ∈ An,0 denote the determinant of the Jacobian matrix of F with respect to the basis {∂/∂x1, …, ∂/∂xn}. Finally, let [JF] ∈ BX denote the equivalence class of JF, modulo IX. Using ∗ to denote multiplication in BX one is able to define a non-degenerate bilinear form β as follows:[2][4]
where is any linear function such that
As mentioned: the signature of β is exactly the index of X at 0.
Example
Consider the case n = 2 of a vector field on the plane. Consider the case where X is given by
Clearly X has an algebraically isolated singularity at 0 since X = 0 if and only if x = y = 0. The ideal IX is given by (x3 − 3xy2, 3x2y − y3), and
The first step for finding the non-degenerate, bilinear form β is to calculate the multiplication table of BX; reducing each entry modulo IX. Whence
∗ | 1 | x | y | x2 | xy | y2 | xy2 | y3 | y4 |
---|---|---|---|---|---|---|---|---|---|
1 | 1 | x | y | x2 | xy | y2 | xy2 | y3 | y4 |
x | x | x2 | xy | 3xy3 | y3/3 | xy2 | y4/3 | 0 | 0 |
y | y | xy | y2 | y3/3 | xy2 | y3 | 0 | y4 | 0 |
x2 | x2 | 3xy2 | y3/3 | y4 | 0 | y4/3 | 0 | 0 | 0 |
xy | xy | y3/3 | xy2 | 0 | y4/3 | 0 | 0 | 0 | 0 |
y2 | y2 | xy2 | y3 | y4/3 | 0 | y4 | 0 | 0 | 0 |
xy2 | xy2 | y4/3 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
y3 | y3 | 0 | y4 | 0 | 0 | 0 | 0 | 0 | 0 |
y4 | y4 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Direct calculation shows that JF = 9(x4 + 2x2y2 + y4), and so [JF] = 24y4. Next one assigns values for . One may take
This choice was made so that as was required by the hypothesis, and to make the calculations involve integers, as opposed to fractions. Applying this to the multiplication table gives the matrix representation of the bilinear form β with respect to the given basis:
The eigenvalues of this matrix are −3, −3, −1, 1, 1, 2, 3, 3 and 4 There are 3 negative eigenvalues (#N = 3), and six positive eigenvalues (#P = 6); meaning that the signature of β is #P − #N = 6 − 3 = +3. It follows that X has Poincaré–Hopf index +3 at the origin.
Topological verification
With this particular choice of X it is possible to verify the Poincaré–Hopf index is +3 by a direct application of the definition of Poincaré–Hopf index.[6] This is very rarely the case, and was the reason for the choice of example. If one takes polar coordinates on the plane, i.e. x = r cos(θ) and y = r sin(θ) then x3 − 3xy2 = r3cos(3θ) and 3x2y − y3 = r3sin(3θ). Restrict X to a circle, centre 0, radius 0 < ε ≪ 1, denoted by C0,ε; and consider the map G : C0,ε → C0,1 given by
The Poincaré–Hopf index of X is, by definition, the topological degree of the map G.[6] Restricting X to the circle C0,ε, for arbitrarily small ε, gives
meaning that as θ makes one rotation about the circle C0,ε in an anti-clockwise direction; the image G(θ) makes three complete, anti-clockwise rotations about the unit circle C0,1. Meaning that the topological degree of G is +3 and that the Poincaré–Hopf index of X at 0 is +3.[6]
References
- ^ Arnold, Vladimir I.; Varchenko, Alexander N.; Gusein-Zade, Sabir M. (2009). Singularities of differentiable maps. Vol. I. The classification of critical points, caustics and wave fronts. Monographs in Mathematics. Vol. 82. Translated by Ian Porteous and Mark Reynolds. Boston, MA: Birkhäuser. p. 84. doi:10.1007/978-1-4612-5154-5. ISBN 0-8176-3187-9. MR 0777682.
- ^ a b c d Brasselet, Jean-Paul; Seade, José; Suwa, Tatsuo (2009), Vector fields on singular varieties, Berlin: Springer, pp. 123–125, doi:10.1007/978-3-642-05205-7, ISBN 978-3-642-05204-0, MR 2574165
- ^ Arnold, Vladimir I. (1978). "The index of a singular point of a vector field, the Petrovskiĭ-Oleĭnik inequalities, and mixed Hodge structures". Functional Analysis and Its Applications. 12 (1): 1–12. doi:10.1007/BF01077558. MR 0498592.
- ^ a b c Gómex Mont, Xavier; Mardešić, Pavao (1997). "The index of a vector field tangent to a hypersurface and the signature of the relative Jacobian determinant". Annales de l'Institut Fourier. 5 (47): 1523–1539. MR 1600363.
- ^ Eisenbud, David; Levine, Harold I. (1977). "An algebraic formula for the degree of a C∞ map germ". Annals of Mathematics. 106 (1): 19–38. doi:10.2307/1971156. JSTOR 1971156. MR 0467800.
- ^ a b c Milnor, John W. (1997), Topology from the Differentiable Viewpoint, Princeton University Press, ISBN 978-0-691-04833-8