• Source: Smooth coarea formula
  • In Riemannian geometry, the smooth coarea formulas relate integrals over the domain of certain mappings with integrals over their codomains.
    Let




    M
    ,

    N



    {\displaystyle \scriptstyle M,\,N}

    be smooth Riemannian manifolds of respective dimensions




    m



    n



    {\displaystyle \scriptstyle m\,\geq \,n}

    . Let




    F
    :
    M



    N



    {\displaystyle \scriptstyle F:M\,\longrightarrow \,N}

    be a smooth surjection such that the pushforward (differential) of




    F



    {\displaystyle \scriptstyle F}

    is surjective almost everywhere. Let




    φ
    :
    M



    [
    0
    ,

    )



    {\displaystyle \scriptstyle \varphi :M\,\longrightarrow \,[0,\infty )}

    a measurable function. Then, the following two equalities hold:







    x

    M


    φ
    (
    x
    )

    d
    M
    =



    y

    N





    x


    F


    1


    (
    y
    )


    φ
    (
    x
    )


    1

    N

    J

    F
    (
    x
    )




    d

    F


    1


    (
    y
    )

    d
    N


    {\displaystyle \int _{x\in M}\varphi (x)\,dM=\int _{y\in N}\int _{x\in F^{-1}(y)}\varphi (x){\frac {1}{N\!J\;F(x)}}\,dF^{-1}(y)\,dN}








    x

    M


    φ
    (
    x
    )
    N

    J

    F
    (
    x
    )

    d
    M
    =



    y

    N





    x


    F


    1


    (
    y
    )


    φ
    (
    x
    )

    d

    F


    1


    (
    y
    )

    d
    N


    {\displaystyle \int _{x\in M}\varphi (x)N\!J\;F(x)\,dM=\int _{y\in N}\int _{x\in F^{-1}(y)}\varphi (x)\,dF^{-1}(y)\,dN}


    where




    N

    J

    F
    (
    x
    )



    {\displaystyle \scriptstyle N\!J\;F(x)}

    is the normal Jacobian of




    F



    {\displaystyle \scriptstyle F}

    , i.e. the determinant of the derivative restricted to the orthogonal complement of its kernel.
    Note that from Sard's lemma almost every point




    y



    N



    {\displaystyle \scriptstyle y\,\in \,N}

    is a regular point of




    F



    {\displaystyle \scriptstyle F}

    and hence the set





    F


    1


    (
    y
    )



    {\displaystyle \scriptstyle F^{-1}(y)}

    is a Riemannian submanifold of




    M



    {\displaystyle \scriptstyle M}

    , so the integrals in the right-hand side of the formulas above make sense.


    References


    Chavel, Isaac (2006) Riemannian Geometry. A Modern Introduction. Second Edition.

Kata Kunci Pencarian: