Poincaré–Hopf theorem

<h2 id="formal-statement">Formal statement</h2>
<p>Let 
  
    
      
        M
      
    
    {\displaystyle M}
  
 be a differentiable manifold, of dimension 
  
    
      
        n
      
    
    {\displaystyle n}
  
, and 
  
    
      
        v
      
    
    {\displaystyle v}
  
 a vector field on 
  
    
      
        M
      
    
    {\displaystyle M}
  
. Suppose that 
  
    
      
        x
      
    
    {\displaystyle x}
  
 is an isolated zero of 
  
    
      
        v
      
    
    {\displaystyle v}
  
, and fix some <a href="/facts/Local_coordinates/cHmdCEYD">local coordinates</a> near 
  
    
      
        x
      
    
    {\displaystyle x}
  
. Pick a closed ball 
  
    
      
        D
      
    
    {\displaystyle D}
  
 centered at 
  
    
      
        x
      
    
    {\displaystyle x}
  
, so that 
  
    
      
        x
      
    
    {\displaystyle x}
  
 is the only zero of 
  
    
      
        v
      
    
    {\displaystyle v}
  
 in 
  
    
      
        D
      
    
    {\displaystyle D}
  
. Then the <a href="/facts/Vector_field/ccFNuPPp">index</a> of 
  
    
      
        v
      
    
    {\displaystyle v}
  
 at 
  
    
      
        x
      
    
    {\displaystyle x}
  
, 
  
    
      
        
          index
          
            x
          
        
        ⁡
        (
        v
        )
      
    
    {\displaystyle \operatorname {index} _{x}(v)}
  
, can be defined as the <a href="/facts/Degree_of_a_continuous_mapping/yWxq9e4B">degree</a> of the map 
  
    
      
        u
        :
        ∂
        D
        →
        
          
            S
          
          
            n
            −
            1
          
        
      
    
    {\displaystyle u:\partial D\to \mathbb {S} ^{n-1}}
  
 from the <a href="/facts/Manifold/rXPERJtu">boundary</a> of 
  
    
      
        D
      
    
    {\displaystyle D}
  
 to the 
  
    
      
        (
        n
        −
        1
        )
      
    
    {\displaystyle (n-1)}
  
-sphere given by  
  
    
      
        u
        (
        z
        )
        =
        v
        (
        z
        )
        
          /
        
        ‖
        v
        (
        z
        )
        ‖
      
    
    {\displaystyle u(z)=v(z)/\|v(z)\|}
  
.
</p><p>Theorem. Let 
  
    
      
        M
      
    
    {\displaystyle M}
  
 be a <a href="/facts/Compact_space/d0cgXJH7">compact</a> <a href="/facts/Differentiable_manifold/aSnbXKcV">differentiable manifold</a>.  Let 
  
    
      
        v
      
    
    {\displaystyle v}
  
 be a <a href="/facts/Vector_field/ccFNuPPp">vector field</a> on 
  
    
      
        M
      
    
    {\displaystyle M}
  
 with isolated zeroes. If 
  
    
      
        M
      
    
    {\displaystyle M}
  
 has <a href="/facts/Manifold_with_boundary/rXPERJtu">boundary</a>, then we insist that 
  
    
      
        v
      
    
    {\displaystyle v}
  
 be pointing in the outward normal direction along the boundary.  Then we have the formula
</p>

∑
          
            i
          
        
        
          index
          
            
              x
              
                i
              
            
          
        
        ⁡
        (
        v
        )
        =
        χ
        (
        M
        )
        
      
    
    {\displaystyle \sum _{i}\operatorname {index} _{x_{i}}(v)=\chi (M)\,}

<p>where the sum of the indices is over all the isolated zeroes of 
  
    
      
        v
      
    
    {\displaystyle v}
  
 and 
  
    
      
        χ
        (
        M
        )
      
    
    {\displaystyle \chi (M)}
  
 is the <a href="/facts/Euler_characteristic/JCjzWUr2">Euler characteristic</a> of 
  
    
      
        M
      
    
    {\displaystyle M}
  
. A particularly useful corollary is when there is a non-vanishing vector field implying Euler characteristic 0.
</p><p>The theorem was proven for two dimensions by <a href="/facts/Henri_Poincar%C3%A9/AazL2j4X">Henri Poincaré</a><a class="footnote-ref" id="fnref:1" href="#fn:1"><sup>1</sup></a> and later generalized to higher dimensions by <a href="/facts/Heinz_Hopf/PMks10NV">Heinz Hopf</a>.<a class="footnote-ref" id="fnref:2" href="#fn:2"><sup>2</sup></a> 
</p>
<h2 id="significance">Significance</h2>
<p>The Euler characteristic of a closed surface is a purely <a href="/facts/Topology/2EqW47cU">topological</a> concept, whereas the index of a vector field is purely <a href="/facts/Analysis_(mathematics)/XXeR26Ih">analytic</a>. Thus, this theorem establishes a deep link between two seemingly unrelated areas of mathematics. It is perhaps as interesting that the proof of this theorem relies heavily on <a href="/facts/Integral/V7lyV997">integration</a>, and, in particular, <a href="/facts/Stokes%27_theorem/z0pAoj2x">Stokes' theorem</a>, which states that the integral of the <a href="/facts/Exterior_derivative/p7HA0wze">exterior derivative</a> of a <a href="/facts/Differential_form/T9gyHtMv">differential form</a> is equal to the integral of that form over the boundary. In the special case of a <a href="/facts/Manifold_(mathematics)/rXPERJtu">manifold</a> without boundary, this amounts to saying that the integral is 0. But by examining vector fields in a sufficiently small neighborhood of a source or sink, we see that sources and sinks contribute <a href="/facts/Integer/PUwwrawx">integer</a> amounts (known as the index) to the total, and they must all sum to 0. This result may be considered[<i>by whom?</i>] one of the earliest of a whole series of theorems (e.g. <a href="/facts/Atiyah%E2%80%93Singer_index_theorem/qwcSxamI">Atiyah–Singer index theorem</a>, <a href="/facts/De_Rham%27s_theorem/4qGKeJq0">De Rham's theorem</a>, <a href="/facts/Grothendieck%E2%80%93Riemann%E2%80%93Roch_theorem/hh5nyYA9">Grothendieck–Riemann–Roch theorem</a>) establishing deep relationships between <a href="/facts/Geometry/wTQf5ffQ">geometric</a> and <a href="/facts/Analysis_(mathematics)/XXeR26Ih">analytical</a> or <a href="/facts/Physics/a7aH2y08">physical</a> concepts. They play an important role in the modern study of both fields.
</p>
<h2 id="sketch-of-proof">Sketch of proof</h2>
<ol><li>Embed <i>M</i> in some high-dimensional Euclidean space. (Use the <a href="/facts/Whitney_embedding_theorem/CyFVXalS">Whitney embedding theorem</a>.)</li>
<li>Take a small neighborhood of <i>M</i> in that Euclidean space, <i>N</i>ε. Extend the vector field to this neighborhood so that it still has the same zeroes and the zeroes have the same indices. In addition, make sure that the extended vector field at the boundary of <i>N</i>ε is directed outwards.</li>
<li>The sum of indices of the zeroes of the old (and new) vector field is equal to the degree of the <a href="/facts/Gauss_map/FHfEvkbk">Gauss map</a> from the boundary of <i>N</i>ε to the (<i>n</i>–1)-dimensional sphere. Thus, the sum of the indices is independent of the actual vector field, and depends only on the manifold <i>M</i>. Technique: cut away all zeroes of the vector field with small neighborhoods. Then use the fact that the degree of a map from the boundary of an n-dimensional manifold to an (<i>n</i>–1)-dimensional sphere, that can be extended to the whole n-dimensional manifold, is zero.</li>
<li>Finally, identify this sum of indices as the Euler characteristic of <i>M</i>. To do that, construct a very specific vector field on <i>M</i> using a <a href="/facts/Triangulation_(topology)/u58jNL6L">triangulation</a> of <i>M</i> for which it is clear that the sum of indices is equal to the Euler characteristic.</li></ol>
<h2 id="generalization">Generalization</h2>
<p>It is still possible to define the index for a vector field with nonisolated zeroes. A construction of this index and the extension of Poincaré–Hopf theorem for vector fields with nonisolated zeroes is outlined in Section 1.1.2 of (Brasselet, Seade & Suwa 2009).
</p><p>Another generalization that use only compact triangulable space and continuous mappings with finitely many fixed points is <a href="/facts/Lefschetz_fixed-point_theorem/TEjN2DQ8">Lefschetz-Hopf theorem</a>.
Since every vector field induce flow on manifold and fixed points of small flows corresponds to zeroes of vector field (and indices of zeroes equals indices of fixed points), then Poincare-Hopf theorem follows immediately from it.
</p>
<h2 id="see-also">See also</h2>
<ul><li><a href="/facts/Eisenbud%E2%80%93Levine%E2%80%93Khimshiashvili_signature_formula/U28C9vhX">Eisenbud–Levine–Khimshiashvili signature formula</a></li>
<li><a href="/facts/Hopf_theorem/ajIhL09K">Hopf theorem</a></li></ul>

<ul><li><a href="https://www.encyclopediaofmath.org/index.php?title=Poincaré–Hopf_theorem">"Poincaré–Hopf theorem"</a>, <i><a href="/facts/Encyclopedia_of_Mathematics/WC6mGtPm">Encyclopedia of Mathematics</a></i>, <a href="/facts/European_Mathematical_Society/B3h7b672">EMS Press</a>, 2001 [1994]</li>
<li>Brasselet, Jean-Paul; Seade, José; Suwa, Tatsuo (2009). <i>Vector fields on singular varieties</i>. Heidelberg: Springer. <a href="/facts/ISBN_(identifier)/15AdSPa9">ISBN</a> 978-3-642-05205-7.</li></ul>

<h2 id="references">References</h2>

<ol>
<li id="fn:1"><p>Henri Poincaré, On curves defined by differential equations (1881–1882) <a href="#fnref:1" class="footnote-back-ref">↩</a></p></li>
<li id="fn:2"><p>H. Hopf, Vektorfelder in n-dimensionalen Mannigfaltigkeiten, Math. Ann. 96 (1926), pp. 209–221. <a href="#fnref:2" class="footnote-back-ref">↩</a></p></li>
</ol>

Poincaré–Hopf theorem open-in-new

Poincaré–Hopf theorem