Frobenius theorem (differential topology)

<h2 id="introduction">Introduction</h2>
<h3>One-form version</h3>
Suppose we are to find the trajectory of a particle in a subset of 3D space, but we do not know its trajectory formula. Instead, we know only that its trajectory satisfies 
 
 
 
 a
 d
 x
 +
 b
 d
 y
 +
 c
 d
 z
 =
 0
 
 
 {\displaystyle adx+bdy+cdz=0}
 
, where 
 
 
 
 a
 ,
 b
 ,
 c
 
 
 {\displaystyle a,b,c}
 
 are smooth functions of 
 
 
 
 (
 x
 ,
 y
 ,
 z
 )
 
 
 {\displaystyle (x,y,z)}
 
. Thus, our only certainty is that if at some moment in time the particle is at location 
 
 
 
 (
 
 x
 
 0
 
 
 ,
 
 y
 
 0
 
 
 ,
 
 z
 
 0
 
 
 )
 
 
 {\displaystyle (x_{0},y_{0},z_{0})}
 
, then its velocity at that moment is restricted within the plane with equation 
 
 
 
 a
 (
 
 x
 
 0
 
 
 ,
 
 y
 
 0
 
 
 ,
 
 z
 
 0
 
 
 )
 [
 x
 −
 
 x
 
 0
 
 
 ]
 +
 b
 (
 
 x
 
 0
 
 
 ,
 
 y
 
 0
 
 
 ,
 
 z
 
 0
 
 
 )
 [
 y
 −
 
 y
 
 0
 
 
 ]
 +
 c
 (
 
 x
 
 0
 
 
 ,
 
 y
 
 0
 
 
 ,
 
 z
 
 0
 
 
 )
 [
 z
 −
 
 z
 
 0
 
 
 ]
 =
 0
 
 
 {\displaystyle a(x_{0},y_{0},z_{0})[x-x_{0}]+b(x_{0},y_{0},z_{0})[y-y_{0}]+c(x_{0},y_{0},z_{0})[z-z_{0}]=0}

In other words, we can draw a "local plane" at each point in 3D space, and we know that the particle's trajectory must be tangent to the local plane at all times.
If we have two equations
 
 
 
 
 
 {
 
 
 
 a
 d
 x
 +
 b
 d
 y
 +
 c
 d
 z
 =
 0
 
 
 
 
 
 a
 ′
 
 d
 x
 +
 
 b
 ′
 
 d
 y
 +
 
 c
 ′
 
 d
 z
 =
 0
 
 
 
 
 
 
 
 
 {\displaystyle {\begin{cases}adx+bdy+cdz=0\\a'dx+b'dy+c'dz=0\end{cases}}}
 
then we can draw two local planes at each point, and their intersection is generically a line, allowing us to uniquely solve for the curve starting at any point. In other words, with two 1-forms, we can <a href="/facts/Foliation/cxml9vyJ">foliate</a> the domain into curves.

If we have only one equation 
 
 
 
 a
 d
 x
 +
 b
 d
 y
 +
 c
 d
 z
 =
 0
 
 
 {\displaystyle adx+bdy+cdz=0}
 
, then we might be able to foliate 
 
 
 
 
 
 R
 
 
 3
 
 
 
 
 {\displaystyle \mathbb {R} ^{3}}
 
 into surfaces, in which case, we can be sure that a curve starting at a certain surface must be restricted to wander within that surface. If not, then a curve starting at any point might end up at any other point in 
 
 
 
 
 
 R
 
 
 3
 
 
 
 
 {\displaystyle \mathbb {R} ^{3}}
 
. One can imagine starting with a cloud of little planes, and <a href="/facts/Quilting/gyhPr0Hx">quilting</a> them together to form a full surface. The main danger is that, if we quilt the little planes two at a time, we might go on a cycle and return to where we began, but shifted by a small amount. If this happens, then we would not get a 2-dimensional surface, but a 3-dimensional blob. An example is shown in the diagram on the right.
If the one-form is integrable, then loops exactly close upon themselves, and each surface would be 2-dimensional. Frobenius' theorem states that this happens precisely when 
 
 
 
 ω
 ∧
 d
 ω
 =
 0
 
 
 {\displaystyle \omega \wedge d\omega =0}
 
 over all of the domain, where 
 
 
 
 ω
 :=
 a
 d
 x
 +
 b
 d
 y
 +
 c
 d
 z
 
 
 {\displaystyle \omega :=adx+bdy+cdz}
 
. The notation is defined in the article on <a href="/facts/One-form_(differential_geometry)/NQLbWmtD">one-forms</a>.
During his development of axiomatic thermodynamics, Carathéodory proved that if 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
 is an integrable one-form on an open subset of 
 
 
 
 
 
 R
 
 
 n
 
 
 
 
 {\displaystyle \mathbb {R} ^{n}}
 
, then 
 
 
 
 ω
 =
 f
 d
 g
 
 
 {\displaystyle \omega =fdg}
 
 for some scalar functions 
 
 
 
 f
 ,
 g
 
 
 {\displaystyle f,g}
 
 on the subset. This is usually called Carathéodory's theorem in axiomatic thermodynamics.<a class="footnote-ref" id="fnref:1" href="#fn:1">1</a><a class="footnote-ref" id="fnref:2" href="#fn:2">2</a> One can prove this intuitively by first constructing the little planes according to 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
, quilting them together into a foliation, then assigning each surface in the foliation with a scalar label. Now for each point 
 
 
 
 p
 
 
 {\displaystyle p}
 
, define 
 
 
 
 g
 (
 p
 )
 
 
 {\displaystyle g(p)}
 
 to be the scalar label of the surface containing point 
 
 
 
 p
 
 
 {\displaystyle p}
 
.

Now, 
 
 
 
 d
 g
 
 
 {\displaystyle dg}
 
 is a one-form that has exactly the same planes as 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
. However, it has "even thickness" everywhere, while 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
 might have "uneven thickness". This can be fixed by a scalar scaling by 
 
 
 
 f
 
 
 {\displaystyle f}
 
, giving 
 
 
 
 ω
 =
 f
 d
 g
 
 
 {\displaystyle \omega =fdg}
 
. This is illustrated on the right.

<h3>Multiple one-forms</h3>
In its most elementary form, the theorem addresses the problem of finding a maximal set of independent solutions of a regular system of first-order linear homogeneous <a href="/facts/Partial_differential_equation/DyPsBlv7">partial differential equations</a>. Let

{
          
            
              f
              
                k
              
              
                i
              
            
            :
            
              
                R
              
              
                n
              
            
            →
            
              R
            
             
            :
             
            1
            ≤
            i
            ≤
            n
            ,
            1
            ≤
            k
            ≤
            r
          
          }
        
      
    
    {\displaystyle \left\{f_{k}^{i}:\mathbf {R} ^{n}\to \mathbf {R} \ :\ 1\leq i\leq n,1\leq k\leq r\right\}}

be a collection of <a href="/facts/Continuously_differentiable/jQqTgLk1">C1</a> functions, with r < n, and such that the matrix ( f ik ) has <a href="/facts/Rank_of_a_matrix/Yp9b5LK3">rank</a> r when evaluated at any point of Rn. Consider the following system of partial differential equations for a C2 function u : Rn → R:

(
        1
        )
        
        
          
            {
            
              
                
                  
                    L
                    
                      1
                    
                  
                  u
                   
                  
                    
                      
                        
                          =
                        
                        
                          
                            d
                            e
                            f
                          
                        
                      
                    
                  
                   
                  
                    ∑
                    
                      i
                    
                  
                  
                    f
                    
                      1
                    
                    
                      i
                    
                  
                  (
                  x
                  )
                  
                    
                      
                        ∂
                        u
                      
                      
                        ∂
                        
                          x
                          
                            i
                          
                        
                      
                    
                  
                  =
                  
                    
                      
                        
                          f
                          →
                        
                      
                    
                    
                      1
                    
                  
                  ⋅
                  ∇
                  u
                  =
                  0
                
              
              
                
                  
                    L
                    
                      2
                    
                  
                  u
                   
                  
                    
                      
                        
                          =
                        
                        
                          
                            d
                            e
                            f
                          
                        
                      
                    
                  
                   
                  
                    ∑
                    
                      i
                    
                  
                  
                    f
                    
                      2
                    
                    
                      i
                    
                  
                  (
                  x
                  )
                  
                    
                      
                        ∂
                        u
                      
                      
                        ∂
                        
                          x
                          
                            i
                          
                        
                      
                    
                  
                  =
                  
                    
                      
                        
                          f
                          →
                        
                      
                    
                    
                      2
                    
                  
                  ⋅
                  ∇
                  u
                  =
                  0
                
              
              
                
                  
                  ⋯
                
              
              
                
                  
                    L
                    
                      r
                    
                  
                  u
                   
                  
                    
                      
                        
                          =
                        
                        
                          
                            d
                            e
                            f
                          
                        
                      
                    
                  
                   
                  
                    ∑
                    
                      i
                    
                  
                  
                    f
                    
                      r
                    
                    
                      i
                    
                  
                  (
                  x
                  )
                  
                    
                      
                        ∂
                        u
                      
                      
                        ∂
                        
                          x
                          
                            i
                          
                        
                      
                    
                  
                  =
                  
                    
                      
                        
                          f
                          →
                        
                      
                    
                    
                      r
                    
                  
                  ⋅
                  ∇
                  u
                  =
                  0
                
              
            
            
          
        
      
    
    {\displaystyle (1)\quad {\begin{cases}L_{1}u\ {\stackrel {\mathrm {def} }{=}}\ \sum _{i}f_{1}^{i}(x){\frac {\partial u}{\partial x^{i}}}={\vec {f}}_{1}\cdot \nabla u=0\\L_{2}u\ {\stackrel {\mathrm {def} }{=}}\ \sum _{i}f_{2}^{i}(x){\frac {\partial u}{\partial x^{i}}}={\vec {f}}_{2}\cdot \nabla u=0\\\qquad \cdots \\L_{r}u\ {\stackrel {\mathrm {def} }{=}}\ \sum _{i}f_{r}^{i}(x){\frac {\partial u}{\partial x^{i}}}={\vec {f}}_{r}\cdot \nabla u=0\end{cases}}}

One seeks conditions on the existence of a collection of solutions u1, ..., un−r such that the gradients ∇u1, ..., ∇un−r are <a href="/facts/Linearly_independent/8JrHfIIa">linearly independent</a>.
The Frobenius theorem asserts that this problem admits a solution locally<a class="footnote-ref" id="fnref:3" href="#fn:3">3</a> if, and only if, the operators Lk satisfy a certain <a href="/facts/Integrability_condition/PKp57zwz">integrability condition</a> known as involutivity. Specifically, they must satisfy relations of the form

L
          
            i
          
        
        
          L
          
            j
          
        
        u
        (
        x
        )
        −
        
          L
          
            j
          
        
        
          L
          
            i
          
        
        u
        (
        x
        )
        =
        
          ∑
          
            k
          
        
        
          c
          
            i
            j
          
          
            k
          
        
        (
        x
        )
        
          L
          
            k
          
        
        u
        (
        x
        )
      
    
    {\displaystyle L_{i}L_{j}u(x)-L_{j}L_{i}u(x)=\sum _{k}c_{ij}^{k}(x)L_{k}u(x)}

for 1 ≤ i, j ≤ r, and all C2 functions u, and for some coefficients ckij(x) that are allowed to depend on x. In other words, the <a href="/facts/Commutator/eYxzFmFY">commutators</a> [Li, Lj] must lie in the <a href="/facts/Linear_span/vPjclEtb">linear span</a> of the Lk at every point. The involutivity condition is a generalization of the commutativity of partial derivatives. In fact, the strategy of proof of the Frobenius theorem is to form linear combinations among the operators Li so that the resulting operators do commute, and then to show that there is a <a href="/facts/Coordinate_system/Mtd0q4SP">coordinate system</a> yi for which these are precisely the partial derivatives with respect to y1, ..., yr.

<h3>From analysis to geometry</h3>
Even though the system is overdetermined there are typically infinitely many solutions. For example, the system of differential equations

{
            
              
                
                  
                    
                      
                        ∂
                        f
                      
                      
                        ∂
                        x
                      
                    
                  
                  +
                  
                    
                      
                        ∂
                        f
                      
                      
                        ∂
                        y
                      
                    
                  
                  =
                  0
                
              
              
                
                  
                    
                      
                        ∂
                        f
                      
                      
                        ∂
                        y
                      
                    
                  
                  +
                  
                    
                      
                        ∂
                        f
                      
                      
                        ∂
                        z
                      
                    
                  
                  =
                  0
                
              
            
            
          
        
      
    
    {\displaystyle {\begin{cases}{\frac {\partial f}{\partial x}}+{\frac {\partial f}{\partial y}}=0\\{\frac {\partial f}{\partial y}}+{\frac {\partial f}{\partial z}}=0\end{cases}}}

clearly permits multiple solutions. Nevertheless, these solutions still have enough structure that they may be completely described. The first observation is that, even if f1 and f2 are two different solutions, the <a href="/facts/Level_surface/JUmNbdNh">level surfaces</a> of f1 and f2 must overlap. In fact, the level surfaces for this system are all planes in R3 of the form x − y + z = C, for C a constant. The second observation is that, once the level surfaces are known, all solutions can then be given in terms of an arbitrary function. Since the value of a solution f on a level surface is constant by definition, define a function C(t) by:

f
        (
        x
        ,
        y
        ,
        z
        )
        =
        C
        (
        t
        )
        
           whenever 
        
        x
        −
        y
        +
        z
        =
        t
        .
      
    
    {\displaystyle f(x,y,z)=C(t){\text{ whenever }}x-y+z=t.}

Conversely, if a function C(t) is given, then each function f given by this expression is a solution of the original equation. Thus, because of the existence of a family of level surfaces, solutions of the original equation are in a one-to-one correspondence with arbitrary functions of one variable.
Frobenius' theorem allows one to establish a similar such correspondence for the more general case of solutions of (1). Suppose that u1, ..., un−r are solutions of the problem (1) satisfying the independence condition on the gradients. Consider the <a href="/facts/Level_set/JUmNbdNh">level sets</a><a class="footnote-ref" id="fnref:4" href="#fn:4">4</a> of (u1, ..., un−r) as functions with values in Rn−r. If v1, ..., vn−r is another such collection of solutions, one can show (using some <a href="/facts/Linear_algebra/8VaB4VWk">linear algebra</a> and the <a href="/facts/Mean_value_theorem/JnjkzQb4">mean value theorem</a>) that this has the same family of level sets but with a possibly different choice of constants for each set. Thus, even though the independent solutions of (1) are not unique, the equation (1) nonetheless determines a unique family of level sets. Just as in the case of the example, general solutions u of (1) are in a one-to-one correspondence with (continuously differentiable) functions on the family of level sets.<a class="footnote-ref" id="fnref:5" href="#fn:5">5</a>
The level sets corresponding to the maximal independent solution sets of (1) are called the integral manifolds because functions on the collection of all integral manifolds correspond in some sense to <a href="/facts/Constant_of_integration/dFzDzFhl">constants of integration</a>. Once one of these constants of integration is known, then the corresponding solution is also known.

<h2 id="frobenius-theorem-in-modern-language">Frobenius' theorem in modern language</h2>
The Frobenius theorem can be restated more economically in modern language. Frobenius' original version of the theorem was stated in terms of <a href="/facts/Pfaffian_system/PKp57zwz">Pfaffian systems</a>, which today can be translated into the language of <a href="/facts/Differential_form/T9gyHtMv">differential forms</a>. An alternative formulation, which is somewhat more intuitive, uses <a href="/facts/Vector_field/ccFNuPPp">vector fields</a>.

<h3>Formulation using vector fields</h3>
In the vector field formulation, the theorem states that a <a href="/facts/Subbundle/KFWrEleF">subbundle</a> of the <a href="/facts/Tangent_bundle/dq8zW9Kb">tangent bundle</a> of a <a href="/facts/Manifold/rXPERJtu">manifold</a> is integrable (or involutive) if and only if it arises from a <a href="/facts/Regular_foliation/cxml9vyJ">regular foliation</a>. In this context, the Frobenius theorem relates <a href="/facts/Integrability_conditions_for_differential_systems/PKp57zwz">integrability</a> to foliation; to state the theorem, both concepts must be clearly defined.
One begins by noting that an arbitrary smooth <a href="/facts/Vector_field/ccFNuPPp">vector field</a> 
 
 
 
 X
 
 
 {\displaystyle X}
 
 on a manifold 
 
 
 
 M
 
 
 {\displaystyle M}
 
 defines a family of <a href="/facts/Curve/xBQ0pTBW">curves</a>, its integral curves 
 
 
 
 u
 :
 I
 →
 M
 
 
 {\displaystyle u:I\to M}
 
 (for intervals 
 
 
 
 I
 
 
 {\displaystyle I}
 
). These are the solutions of 
 
 
 
 
 
 
 u
 ˙
 
 
 
 (
 t
 )
 =
 
 X
 
 u
 (
 t
 )
 
 
 
 
 {\displaystyle {\dot {u}}(t)=X_{u(t)}}
 
, which is a system of first-order <a href="/facts/Ordinary_differential_equation/ygKGQ8kD">ordinary differential equations</a>, whose solvability is guaranteed by the <a href="/facts/Picard%25E2%2580%2593Lindel%25C3%25B6f_theorem/yKHSWoMJ">Picard–Lindelöf theorem</a>. If the vector field 
 
 
 
 X
 
 
 {\displaystyle X}
 
 is nowhere zero then it defines a one-dimensional subbundle of the tangent bundle of 
 
 
 
 M
 
 
 {\displaystyle M}
 
, and the integral curves form a regular foliation of 
 
 
 
 M
 
 
 {\displaystyle M}
 
. Thus, one-dimensional subbundles are always integrable.
If the subbundle has dimension greater than one, a condition needs to be imposed.
One says that a <a href="/facts/Subbundle/KFWrEleF">subbundle</a> 
 
 
 
 E
 ⊂
 T
 M
 
 
 {\displaystyle E\subset TM}
 
 of the <a href="/facts/Tangent_bundle/dq8zW9Kb">tangent bundle</a> 
 
 
 
 T
 M
 
 
 {\displaystyle TM}
 
 is integrable (or involutive), if, for any two vector fields 
 
 
 
 X
 
 
 {\displaystyle X}
 
 and 
 
 
 
 Y
 
 
 {\displaystyle Y}
 
 taking values in 
 
 
 
 E
 
 
 {\displaystyle E}
 
, the <a href="/facts/Lie_bracket_of_vector_fields/n22pPTnV">Lie bracket</a> 
 
 
 
 [
 X
 ,
 Y
 ]
 
 
 {\displaystyle [X,Y]}
 
 takes values in 
 
 
 
 E
 
 
 {\displaystyle E}
 
 as well. This notion of integrability need only be defined locally; that is, the existence of the vector fields 
 
 
 
 X
 
 
 {\displaystyle X}
 
 and 
 
 
 
 Y
 
 
 {\displaystyle Y}
 
 and their integrability need only be defined on subsets of 
 
 
 
 M
 
 
 {\displaystyle M}
 
.
Several definitions of <a href="/facts/Foliation/cxml9vyJ">foliation</a> exist. Here we use the following:
Definition. A p-dimensional, class Cr foliation of an n-dimensional manifold M is a decomposition of M into a union of <a href="/facts/Disjoint_sets/nLr8XRVd">disjoint</a> connected submanifolds {Lα}α∈A, called the leaves of the foliation, with the following property: Every point in M has a neighborhood U and a system of local, class Cr coordinates x=(x1, ⋅⋅⋅, xn) : U→Rn such that for each leaf Lα, the components of U ∩ Lα are described by the equations xp+1=constant, ⋅⋅⋅, xn=constant. A foliation is denoted by 
 
 
 
 
 
 F
 
 
 
 
 {\displaystyle {\mathcal {F}}}
 
={Lα}α∈A.<a class="footnote-ref" id="fnref:6" href="#fn:6">6</a>
Trivially, any <a href="/facts/Foliation/cxml9vyJ">foliation</a> of 
 
 
 
 M
 
 
 {\displaystyle M}
 
 defines an integrable subbundle, since if 
 
 
 
 p
 ∈
 M
 
 
 {\displaystyle p\in M}
 
 and 
 
 
 
 N
 ⊂
 M
 
 
 {\displaystyle N\subset M}
 
 is the leaf of the foliation passing through 
 
 
 
 p
 
 
 {\displaystyle p}
 
 then 
 
 
 
 
 E
 
 p
 
 
 =
 
 T
 
 p
 
 
 N
 
 
 {\displaystyle E_{p}=T_{p}N}
 
 is integrable. Frobenius' theorem states that the converse is also true:
Given the above definitions, Frobenius' theorem states that a subbundle 
 
 
 
 E
 
 
 {\displaystyle E}
 
 is integrable if and only if the subbundle 
 
 
 
 E
 
 
 {\displaystyle E}
 
 arises from a regular foliation of 
 
 
 
 M
 
 
 {\displaystyle M}
 
.

<h3>Differential forms formulation</h3>
Let U be an open set in a manifold M, Ω1(U) be the space of smooth, differentiable <a href="/facts/1-form/NQLbWmtD">1-forms</a> on U, and F be a <a href="/facts/Submodule/jP0LIhFL">submodule</a> of Ω1(U) of <a href="/facts/Rank_(linear_algebra)/Yp9b5LK3">rank</a> r, the rank being constant in value over U. The Frobenius theorem states that F is <a href="/facts/Integrability_conditions_for_differential_systems/PKp57zwz">integrable</a> if and only if for every p in U the <a href="/facts/Stalk_of_a_sheaf/qAD2etdW">stalk</a> Fp is generated by r <a href="/facts/Exact_differential_form/uvEUbzUK">exact differential forms</a>.
Geometrically, the theorem states that an integrable module of 1-forms of rank r is the same thing as a codimension-r <a href="/facts/Foliation/cxml9vyJ">foliation</a>. The correspondence to the definition in terms of vector fields given in the introduction follows from the close relationship between <a href="/facts/Differential_form/T9gyHtMv">differential forms</a> and <a href="/facts/Lie_derivative/1KSJvrzk">Lie derivatives</a>. Frobenius' theorem is one of the basic tools for the study of <a href="/facts/Vector_field/ccFNuPPp">vector fields</a> and foliations.
There are thus two forms of the theorem: one which operates with <a href="/facts/Distribution_(differential_geometry)/N3My5NLR">distributions</a>, that is smooth subbundles D of the tangent bundle TM; and the other which operates with subbundles of the graded ring Ω(M) of all forms on M. These two forms are related by duality. If D is a smooth tangent distribution on M, then the annihilator of D, I(D) consists of all forms 
 
 
 
 α
 ∈
 
 Ω
 
 k
 
 
 (
 M
 )
 
 
 {\displaystyle \alpha \in \Omega ^{k}(M)}
 
 (for any 
 
 
 
 k
 ∈
 {
 1
 ,
 …
 ,
 dim
 ⁡
 M
 }
 
 
 {\displaystyle k\in \{1,\dots ,\operatorname {dim} M\}}
 
) such that

α
        (
        
          v
          
            1
          
        
        ,
        …
        ,
        
          v
          
            k
          
        
        )
        =
        0
      
    
    {\displaystyle \alpha (v_{1},\dots ,v_{k})=0}

for all 
 
 
 
 
 v
 
 1
 
 
 ,
 …
 ,
 
 v
 
 k
 
 
 ∈
 D
 
 
 {\displaystyle v_{1},\dots ,v_{k}\in D}
 
. The set I(D) forms a subring and, in fact, an ideal in Ω(M). Furthermore, using the definition of the <a href="/facts/Exterior_derivative/p7HA0wze">exterior derivative</a>, it can be shown that I(D) is closed under exterior differentiation (it is a <a href="/facts/Differential_ideal/06MCfkLa">differential ideal</a>) if and only if D is involutive. Consequently, the Frobenius theorem takes on the equivalent form that I(D) is closed under exterior differentiation if and only if D is integrable.

<h2 id="generalizations">Generalizations</h2>
The theorem may be generalized in a variety of ways.

<h3>Infinite dimensions</h3>
One infinite-dimensional generalization is as follows.<a class="footnote-ref" id="fnref:7" href="#fn:7">7</a> Let X and Y be <a href="/facts/Banach_space/sDEIk7EC">Banach spaces</a>, and A ⊂ X, B ⊂ Y a pair of <a href="/facts/Open_set/QfTCIqMu">open sets</a>. Let

F
        :
        A
        ×
        B
        →
        L
        (
        X
        ,
        Y
        )
      
    
    {\displaystyle F:A\times B\to L(X,Y)}

be a <a href="/facts/Continuously_differentiable_function/jQqTgLk1">continuously differentiable function</a> of the <a href="/facts/Cartesian_product/BfqBWzdL">Cartesian product</a> (which inherits a <a href="/facts/Differentiable_structure/nHHTJud2">differentiable structure</a> from its inclusion into X ×Y ) into the space L(X,Y) of <a href="/facts/Continuous_linear_transformation/KcqjDBrJ">continuous linear transformations</a> of X into Y. A differentiable mapping u : A → B is a solution of the differential equation

(
        1
        )
        
        
          y
          ′
        
        =
        F
        (
        x
        ,
        y
        )
      
    
    {\displaystyle (1)\quad y'=F(x,y)}

if

∀
        x
        ∈
        A
        :
        
        
          u
          ′
        
        (
        x
        )
        =
        F
        (
        x
        ,
        u
        (
        x
        )
        )
        .
      
    
    {\displaystyle \forall x\in A:\quad u'(x)=F(x,u(x)).}

The equation (1) is completely integrable if for each 
 
 
 
 (
 
 x
 
 0
 
 
 ,
 
 y
 
 0
 
 
 )
 ∈
 A
 ×
 B
 
 
 {\displaystyle (x_{0},y_{0})\in A\times B}
 
, there is a neighborhood U of x0 such that (1) has a unique solution u(x) defined on U such that u(x0)=y0.
The conditions of the Frobenius theorem depend on whether the underlying <a href="/facts/Field_(mathematics)/xAjAS4ko">field</a> is R or C. If it is R, then assume F is continuously differentiable. If it is C, then assume F is twice continuously differentiable. Then (1) is completely integrable at each point of A × B if and only if

D
          
            1
          
        
        F
        (
        x
        ,
        y
        )
        ⋅
        (
        
          s
          
            1
          
        
        ,
        
          s
          
            2
          
        
        )
        +
        
          D
          
            2
          
        
        F
        (
        x
        ,
        y
        )
        ⋅
        (
        F
        (
        x
        ,
        y
        )
        ⋅
        
          s
          
            1
          
        
        ,
        
          s
          
            2
          
        
        )
        =
        
          D
          
            1
          
        
        F
        (
        x
        ,
        y
        )
        ⋅
        (
        
          s
          
            2
          
        
        ,
        
          s
          
            1
          
        
        )
        +
        
          D
          
            2
          
        
        F
        (
        x
        ,
        y
        )
        ⋅
        (
        F
        (
        x
        ,
        y
        )
        ⋅
        
          s
          
            2
          
        
        ,
        
          s
          
            1
          
        
        )
      
    
    {\displaystyle D_{1}F(x,y)\cdot (s_{1},s_{2})+D_{2}F(x,y)\cdot (F(x,y)\cdot s_{1},s_{2})=D_{1}F(x,y)\cdot (s_{2},s_{1})+D_{2}F(x,y)\cdot (F(x,y)\cdot s_{2},s_{1})}

for all s1, s2 ∈ X. Here D1 (resp. D2) denotes the partial derivative with respect to the first (resp. second) variable; the <a href="/facts/Dot_product/tNz8MLjT">dot product</a> denotes the action of the linear operator F(x, y) ∈ L(X, Y), as well as the actions of the operators D1F(x, y) ∈ L(X, L(X, Y)) and D2F(x, y) ∈ L(Y, L(X, Y)).

<h4>Banach manifolds</h4>
The infinite-dimensional version of the Frobenius theorem also holds on <a href="/facts/Banach_manifold/EySHBrBz">Banach manifolds</a>.<a class="footnote-ref" id="fnref:8" href="#fn:8">8</a> The statement is essentially the same as the finite-dimensional version.
Let M be a Banach manifold of class at least C2. Let E be a subbundle of the tangent bundle of M. The bundle E is involutive if, for each point p ∈ M and pair of sections X and Y of E defined in a neighborhood of p, the Lie bracket of X and Y evaluated at p, lies in Ep:

[
        X
        ,
        Y
        
          ]
          
            p
          
        
        ∈
        
          E
          
            p
          
        
      
    
    {\displaystyle [X,Y]_{p}\in E_{p}}

On the other hand, E is integrable if, for each p ∈ M, there is an immersed submanifold φ : N → M whose image contains p, such that the <a href="/facts/Pushforward_(differential)/xLsoLwVH">differential</a> of φ is an isomorphism of TN with φ−1E.
The Frobenius theorem states that a subbundle E is integrable if and only if it is involutive.

<h3>Holomorphic forms</h3>
The statement of the theorem remains true for <a href="/facts/Complex_differential_form/307vM8pd">holomorphic 1-forms</a> on <a href="/facts/Complex_manifold/DlvPnpzm">complex manifolds</a> — manifolds over C with biholomorphic <a href="/facts/Transition_map/pjoya29t">transition functions</a>.<a class="footnote-ref" id="fnref:9" href="#fn:9">9</a>
Specifically, if 
 
 
 
 
 ω
 
 1
 
 
 ,
 …
 ,
 
 ω
 
 r
 
 
 
 
 {\displaystyle \omega ^{1},\dots ,\omega ^{r}}
 
 are r linearly independent holomorphic 1-forms on an open set in Cn such that

d
        
          ω
          
            j
          
        
        =
        
          ∑
          
            i
            =
            1
          
          
            r
          
        
        
          ψ
          
            i
          
          
            j
          
        
        ∧
        
          ω
          
            i
          
        
      
    
    {\displaystyle d\omega ^{j}=\sum _{i=1}^{r}\psi _{i}^{j}\wedge \omega ^{i}}

for some system of holomorphic 1-forms ψ ji, 1 ≤ i, j ≤ r, then there exist holomorphic functions fij and gi such that, on a possibly smaller domain,

ω
          
            j
          
        
        =
        
          ∑
          
            i
            =
            1
          
          
            r
          
        
        
          f
          
            i
          
          
            j
          
        
        d
        
          g
          
            i
          
        
        .
      
    
    {\displaystyle \omega ^{j}=\sum _{i=1}^{r}f_{i}^{j}dg^{i}.}

This result holds locally in the same sense as the other versions of the Frobenius theorem. In particular, the fact that it has been stated for domains in Cn is not restrictive.

<h3>Higher degree forms</h3>
The statement does not generalize to higher degree forms, although there is a number of partial results such as <a href="/facts/Darboux%2527s_theorem/81zX7bxI">Darboux's theorem</a> and the <a href="/facts/Cartan-K%25C3%25A4hler_theorem/3nkGyJbh">Cartan-Kähler theorem</a>.

<h2 id="history">History</h2>
Despite being named for <a href="/facts/Ferdinand_Georg_Frobenius/4cXXQyu5">Ferdinand Georg Frobenius</a>, the theorem was first proven by <a href="/facts/Alfred_Clebsch/NV4IyEAk">Alfred Clebsch</a> and <a href="/facts/Feodor_Deahna/wqm02fSS">Feodor Deahna</a>. Deahna was the first to establish the <a href="/facts/Necessity_and_sufficiency/G45aDCY9">sufficient</a> conditions for the theorem, and Clebsch developed the <a href="/facts/Necessity_and_sufficiency/G45aDCY9">necessary</a> conditions. Frobenius is responsible for applying the theorem to <a href="/facts/Pfaffian_system/PKp57zwz">Pfaffian systems</a>, thus paving the way for its usage in differential topology.

<h2 id="applications">Applications</h2>
<ul><li>In <a href="/facts/Classical_mechanics/ykG8upsN">classical mechanics</a>, the integrability of a system's constraint equations determines whether the system is <a href="/facts/Holonomic_constraints/cvxCH0Ar">holonomic</a> or <a href="/facts/Nonholonomic_system/5VHl7nMU">nonholonomic</a>.</li>
<li>In <a href="/facts/Microeconomic_theory/mX6Ek69Y">microeconomic theory</a>, Frobenius' theorem can be used to prove the existence of a solution to the problem of <a href="/facts/Integrability_of_demand/nA2xTvSR"> integrability of demand functions</a>.</li></ul>
<h3>Carathéodory's axiomatic thermodynamics</h3>
In <a href="/facts/Classical_thermodynamics/q4hIx3yP">classical thermodynamics</a>, Frobenius' theorem can be used to construct entropy and temperature in Carathéodory's formalism.<a class="footnote-ref" id="fnref:10" href="#fn:10">10</a><a class="footnote-ref" id="fnref:11" href="#fn:11">11</a>
Specifically, Carathéodory considered a thermodynamic system (concretely one can imagine a piston of gas) that can interact with the outside world by either heat conduction (such as setting the piston on fire) or mechanical work (pushing on the piston). He then defined "adiabatic process" as any process that the system may undergo without heat conduction, and defined a relation of "<a href="/facts/Adiabatic_accessibility/L0uKlrlw">adiabatic accessibility</a>" thus: if the system can go from state A to state B after an adiabatic process, then 
 
 
 
 B
 
 
 {\displaystyle B}
 
 is adiabatically accessible from 
 
 
 
 A
 
 
 {\displaystyle A}
 
. Write it as 
 
 
 
 A
 ⪰
 B
 
 
 {\displaystyle A\succeq B}
 
.
Now assume that

<ul><li>For any pair of states 
 
 
 
 A
 ,
 B
 
 
 {\displaystyle A,B}
 
, at least one of 
 
 
 
 A
 ⪰
 B
 
 
 {\displaystyle A\succeq B}
 
 and 
 
 
 
 B
 ⪰
 A
 
 
 {\displaystyle B\succeq A}
 
 holds.</li>
<li>For any state 
 
 
 
 A
 
 
 {\displaystyle A}
 
, and any neighborhood of 
 
 
 
 A
 
 
 {\displaystyle A}
 
, there exists a state 
 
 
 
 B
 
 
 {\displaystyle B}
 
 in the neighborhood, such that 
 
 
 
 B
 
 
 {\displaystyle B}
 
 is adiabatically inaccessible from 
 
 
 
 A
 
 
 {\displaystyle A}
 
.</li></ul>
Then, we can foliate the state space into subsets of states that are mutually adiabatically accessible. With mild assumptions on the smoothness of 
 
 
 
 ⪰
 
 
 {\displaystyle \succeq }
 
, each subset is a manifold of codimension 1. Call these manifolds "adiabatic surfaces".
By the <a href="/facts/First_law_of_thermodynamics/hjL95WCr">first law of thermodynamics</a>, there exists a scalar function 
 
 
 
 U
 
 
 {\displaystyle U}
 
 ("internal energy") on the state space, such that
 
 
 
 d
 U
 =
 δ
 W
 +
 δ
 Q
 =
 
 ∑
 
 i
 
 
 
 X
 
 i
 
 
 d
 
 x
 
 i
 
 
 +
 δ
 Q
 
 
 {\displaystyle dU=\delta W+\delta Q=\sum _{i}X_{i}dx_{i}+\delta Q}
 
where 
 
 
 
 
 X
 
 1
 
 
 d
 
 x
 
 1
 
 
 ,
 .
 .
 .
 ,
 
 X
 
 n
 
 
 d
 
 x
 
 n
 
 
 
 
 {\displaystyle X_{1}dx_{1},...,X_{n}dx_{n}}
 
 are the possible ways to perform mechanical work on the system. For example, if the system is a tank of ideal gas, then 
 
 
 
 δ
 W
 =
 −
 p
 d
 V
 
 
 {\displaystyle \delta W=-pdV}
 
.
Now, define the one-form on the state space
 
 
 
 ω
 :=
 d
 U
 −
 
 ∑
 
 i
 
 
 
 X
 
 i
 
 
 d
 
 x
 
 i
 
 
 
 
 {\displaystyle \omega :=dU-\sum _{i}X_{i}dx_{i}}
 
Now, since the adiabatic surfaces are tangent to 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
 at every point in state space, 
 
 
 
 ω
 
 
 {\displaystyle \omega }
 
 is integrable, so by Carathéodory's theorem, there exists two scalar functions 
 
 
 
 T
 ,
 S
 
 
 {\displaystyle T,S}
 
 on state space, such that 
 
 
 
 ω
 =
 T
 d
 S
 
 
 {\displaystyle \omega =TdS}
 
. These are the temperature and entropy functions, up to a multiplicative constant. 
By plugging in the ideal gas laws, and noting that <a href="/facts/Joule_expansion/kXcheXjn">Joule expansion</a> is an (irreversible) adiabatic process, we can fix the sign of 
 
 
 
 d
 S
 
 
 {\displaystyle dS}
 
, and find that 
 
 
 
 A
 ⪰
 B
 
 
 {\displaystyle A\succeq B}
 
 means 
 
 
 
 S
 (
 A
 )
 ≤
 S
 (
 B
 )
 
 
 {\displaystyle S(A)\leq S(B)}
 
. That is, entropy is preserved in reversible adiabatic processes, and increases during irreversible adiabatic processes.

<h2 id="see-also">See also</h2>
<ul><li><a href="/facts/Integrability_conditions_for_differential_systems/PKp57zwz">Integrability conditions for differential systems</a></li>
<li><a href="/facts/Domain-straightening_theorem/tKWv6x1d">Domain-straightening theorem</a></li>
<li><a href="/facts/Almost_complex_manifold/wsXTsuC6">Newlander-Nirenberg Theorem</a></li></ul>
<h2 id="notes">Notes</h2>

<ul><li><a href="/facts/H._B._Lawson/mcTSg513">Lawson, H.B.</a> (1977). The Qualitative Theory of Foliations. American Mathematical Society CBMS Series. Vol. 27. AMS.</li>
<li><a href="/facts/Ralph_Abraham_(mathematician)/3GwuMpup">Abraham, Ralph.</a>; <a href="/facts/Jerrold_E._Marsden/I07WtXQe">Marsden, Jerrold E.</a> (2008) [1978]. <a href="https://books.google.com/books?id=YAEKBAAAQBAJ&pg=PA93">"2.2.26 Frobenius' Theorem"</a>. Foundations of Mechanics (2nd ed.). American Mathematical Society. p. 93. <a href="/facts/ISBN_(identifier)/15AdSPa9">ISBN</a> 9780821844380.</li>
<li>Clebsch, A. (1866). <a href="https://zenodo.org/record/2050541">"Ueber die simultane Integration linearer partieller Differentialgleichungen"</a>. J. Reine. Angew. Math. (Crelle). 1866 (65): 257–268. <a href="/facts/Doi_(identifier)/muM9Etpq">doi</a>:<a href="https://doi.org/10.1515%2Fcrll.1866.65.257">10.1515/crll.1866.65.257</a>. <a href="/facts/S2CID_(identifier)/ldJsHa2Y">S2CID</a> <a href="https://api.semanticscholar.org/CorpusID:122439486">122439486</a>.</li>
<li>Deahna, F. (1840). <a href="https://zenodo.org/record/2312273">"Über die Bedingungen der Integrabilitat ..."</a> J. Reine Angew. Math. 20: 340–350. <a href="/facts/Doi_(identifier)/muM9Etpq">doi</a>:<a href="https://doi.org/10.1515%2Fcrll.1840.20.340">10.1515/crll.1840.20.340</a>. <a href="/facts/S2CID_(identifier)/ldJsHa2Y">S2CID</a> <a href="https://api.semanticscholar.org/CorpusID:120057555">120057555</a>.</li>
<li>Frobenius, G. (1877). "Über das Pfaffsche Problem". J. Reine Angew. Math. 1877 (82): 230–315. <a href="/facts/Doi_(identifier)/muM9Etpq">doi</a>:<a href="https://doi.org/10.1515%2Fcrll.1877.82.230">10.1515/crll.1877.82.230</a>. <a href="/facts/S2CID_(identifier)/ldJsHa2Y">S2CID</a> <a href="https://api.semanticscholar.org/CorpusID:119848431">119848431</a>.</li></ul>

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

<ol>
<li id="fn:1">Buchdahl, H. A. (April 1949). "On the Unrestricted Theorem of Carathéodory and Its Application in the Treatment of the Second Law of Thermodynamics". American Journal of Physics. 17 (4): 212–218. Bibcode:1949AmJPh..17..212B. doi:10.1119/1.1989552. ISSN 0002-9505. <a href="https://pubs.aip.org/aapt/ajp/article/17/4/212-218/1033957" target="_blank">https://pubs.aip.org/aapt/ajp/article/17/4/212-218/1033957</a> <a href="#fnref:1" class="footnote-back-ref">↩</a></li>
<li id="fn:2">Carathéodory, C. (1909). "Untersuchungen über die Grundlagen der Thermodynamik". Mathematische Annalen. 67 (3): 355–386. doi:10.1007/BF01450409. ISSN 0025-5831. <a href="https://eudml.org/doc/158411" target="_blank">https://eudml.org/doc/158411</a> <a href="#fnref:2" class="footnote-back-ref">↩</a></li>
<li id="fn:3">Here locally means inside small enough open subsets of Rn. Henceforth, when we speak of a solution, we mean a local solution. <a href="#fnref:3" class="footnote-back-ref">↩</a></li>
<li id="fn:4">A level set is a subset of Rn corresponding to the locus of:

(u1, ..., un−r) = (c1, ..., cn−r),

for some constants ci. <a href="#fnref:4" class="footnote-back-ref">↩</a></li>
<li id="fn:5">The notion of a continuously differentiable function on a family of level sets can be made rigorous by means of the implicit function theorem. <a href="/wiki/Implicit_function_theorem" target="_blank">/wiki/Implicit_function_theorem</a> <a href="#fnref:5" class="footnote-back-ref">↩</a></li>
<li id="fn:6">Lawson, H. Blaine (1974), "Foliations", Bulletin of the American Mathematical Society, 80 (3): 369–418, ISSN 0040-9383, Zbl 0293.57014 <a href="/wiki/Bulletin_of_the_American_Mathematical_Society" target="_blank">/wiki/Bulletin_of_the_American_Mathematical_Society</a> <a href="#fnref:6" class="footnote-back-ref">↩</a></li>
<li id="fn:7">Dieudonné, J (1969). "Ch. 10.9". Foundations of modern analysis. Academic Press. ISBN 9780122155307. <a href="9780122155307" target="_blank">9780122155307</a> <a href="#fnref:7" class="footnote-back-ref">↩</a></li>
<li id="fn:8">Lang, S. (1995). "Ch. VI: The theorem of Frobenius". Differential and Riemannian manifolds. Springer-Verlag. ISBN 978-0-387-94338-1. <a href="978-0-387-94338-1" target="_blank">978-0-387-94338-1</a> <a href="#fnref:8" class="footnote-back-ref">↩</a></li>
<li id="fn:9">Kobayashi, S.; Nomizu, K. (2009) [1969]. "Appendix 8". Foundations of Differential Geometry. Wiley Classics Library. Vol. 2. Wiley. ISBN 978-0-471-15732-8. Zbl 0175.48504. <a href="978-0-471-15732-8" target="_blank">978-0-471-15732-8</a> <a href="#fnref:9" class="footnote-back-ref">↩</a></li>
<li id="fn:10">Buchdahl, H. A. (April 1949). "On the Unrestricted Theorem of Carathéodory and Its Application in the Treatment of the Second Law of Thermodynamics". American Journal of Physics. 17 (4): 212–218. Bibcode:1949AmJPh..17..212B. doi:10.1119/1.1989552. ISSN 0002-9505. <a href="https://pubs.aip.org/aapt/ajp/article/17/4/212-218/1033957" target="_blank">https://pubs.aip.org/aapt/ajp/article/17/4/212-218/1033957</a> <a href="#fnref:10" class="footnote-back-ref">↩</a></li>
<li id="fn:11">Buchdahl, H. A. (1960-03-01). "The Concepts of Classical Thermodynamics". American Journal of Physics. 28 (3): 196–201. Bibcode:1960AmJPh..28..196B. doi:10.1119/1.1935102. ISSN 0002-9505. <a href="https://ui.adsabs.harvard.edu/abs/1960AmJPh..28..196B" target="_blank">https://ui.adsabs.harvard.edu/abs/1960AmJPh..28..196B</a> <a href="#fnref:11" class="footnote-back-ref">↩</a></li>
</ol>

Frobenius theorem (differential topology) open-in-new

Frobenius theorem (differential topology)