Young measure

In mathematical analysis, a Young measure is a parameterized measure that is associated with certain subsequences of a given bounded sequence of measurable functions. They are a quantification of the oscillation effect of the sequence in the limit. Young measures have applications in the calculus of variations, especially models from material science, and the study of nonlinear partial differential equations, as well as in various optimization (or optimal control problems). They are named after Laurence Chisholm Young who invented them, already in 1937 in one dimension (curves) and later in higher dimensions in 1942. ^[1]

Young measures provide a solution to Hilbert’s twentieth problem, as a broad class of problems in the calculus of variations have solutions in the form of Young measures. ^[2]

Definition

Intuition

Young constructed the Young measure in order to complete sets of ordinary curves in the calculus of variations. That is, Young measures are "generalized curves". ^[2]

Consider the problem of ${\displaystyle \min _{u}I(u)=\int _{0}^{1}(u'(x)^{2}-1)^{2}+u(x)^{2}dx}$, where ${\displaystyle u}$ is a function such that ${\displaystyle u(0)=u(1)=0}$, and continuously differentiable. It is clear that we should pick ${\displaystyle u}$ to have value close to zero, and its slope close to ${\displaystyle \pm 1}$. That is, the curve should be a tight jagged line hugging close to the x-axis. No function can reach the minimum value of ${\displaystyle I=0}$, but we can construct a sequence of functions ${\displaystyle u_{1},u_{2},\dots }$ that are increasingly jagged, such that ${\displaystyle I(u_{n})\to 0}$.

The pointwise limit ${\displaystyle \lim u_{n}}$ is identically zero, but the pointwise limit ${\displaystyle \lim _{n}u_{n}'}$ does not exist. Instead, it is a fine mist that has half of its weight on ${\displaystyle +1}$, and the other half on ${\displaystyle -1}$.

Suppose that ${\displaystyle F}$ is a functional defined by ${\displaystyle F(u)=\int _{0}^{1}f(t,u(t),u'(t))dt}$, where ${\displaystyle f}$ is continuous, then $${\displaystyle \lim _{n}F(u_{n})={\frac {1}{2}}\int _{0}^{1}f(t,0,-1)dt+{\frac {1}{2}}\int _{0}^{1}f(t,0,+1)dt}$$so in the weak sense, we can define ${\displaystyle \lim _{n}u_{n}}$ to be a "function" whose value is zero and whose derivative is ${\displaystyle {\frac {1}{2}}\delta _{-1}+{\frac {1}{2}}\delta _{+1}}$. In particular, it would mean that ${\displaystyle I(\lim _{n}u_{n})=0}$.

Motivation

The definition of Young measures is motivated by the following theorem: Let m, n be arbitrary positive integers, let ${\displaystyle U}$ be an open bounded subset of ${\displaystyle \mathbb {R} ^{n}}$ and ${\displaystyle \{f_{k}\}_{k=1}^{\infty }}$ be a bounded sequence in ${\displaystyle L^{p}(U,\mathbb {R} ^{m})}$^{[ clarification needed ]}. Then there exists a subsequence ${\displaystyle \{f_{k_{j}}\}_{j=1}^{\infty }\subset \{f_{k}\}_{k=1}^{\infty }}$ and for almost every ${\displaystyle x\in U}$ a Borel probability measure ${\displaystyle \nu _{x}}$ on ${\displaystyle \mathbb {R} ^{m}}$ such that for each ${\displaystyle F\in C(\mathbb {R} ^{m})}$ we have

${\displaystyle F\circ f_{k_{j}}(x){\rightharpoonup }\int _{\mathbb {R} ^{m}}F(y)d\nu _{x}(y)}$

weakly in ${\displaystyle L^{p}(U)}$ if the limit exists (or weakly* in ${\displaystyle L^{\infty }(U)}$ in case of ${\displaystyle p=+\infty }$). The measures ${\displaystyle \nu _{x}}$ are called the Young measures generated by the sequence ${\displaystyle \{f_{k_{j}}\}_{j=1}^{\infty }}$.

A partial converse is also true: If for each ${\displaystyle x\in U}$ we have a Borel measure ${\displaystyle \nu _{x}}$ on ${\displaystyle \mathbb {R} ^{m}}$ such that ${\displaystyle \int _{U}\int _{\mathbb {R} ^{m}}\|y\|^{p}d\nu _{x}(y)dx<+\infty }$, then there exists a sequence ${\displaystyle \{f_{k}\}_{k=1}^{\infty }\subseteq L^{p}(U,\mathbb {R} ^{m})}$, bounded in ${\displaystyle L^{p}(U,\mathbb {R} ^{m})}$, that has the same weak convergence property as above.

More generally, for any Carathéodory function ${\displaystyle G(x,A):U\times R^{m}\to R}$, the limit

${\displaystyle \lim _{j\to \infty }\int _{U}G(x,f_{j}(x))\ dx,}$

if it exists, will be given by ^[3]

${\displaystyle \int _{U}\int _{\mathbb {R} ^{m}}G(x,A)\ d\nu _{x}(A)\ dx}$.

Young's original idea in the case ${\displaystyle G\in C_{0}(U\times \mathbb {R} ^{m})}$ was to consider for each integer ${\displaystyle j\geq 1}$ the uniform measure, let's say ${\displaystyle \Gamma _{j}:=(id,f_{j})_{\sharp }L^{d}\llcorner U,}$ concentrated on graph of the function ${\displaystyle f_{j}.}$ (Here, ${\displaystyle L^{d}\llcorner U}$ is the restriction of the Lebesgue measure on ${\displaystyle U.}$) By taking the weak* limit of these measures as elements of ${\displaystyle C_{0}(U\times \mathbb {R} ^{m})^{\star },}$ we have

${\displaystyle \langle \Gamma _{j},G\rangle =\int _{U}G(x,f_{j}(x))\ dx\to \langle \Gamma ,G\rangle ,}$

where ${\displaystyle \Gamma }$ is the mentioned weak limit. After a disintegration of the measure ${\displaystyle \Gamma }$ on the product space ${\displaystyle \Omega \times \mathbb {R} ^{m},}$ we get the parameterized measure ${\displaystyle \nu _{x}}$.

General definition

Let ${\displaystyle m,n}$ be arbitrary positive integers, let ${\displaystyle U}$ be an open and bounded subset of ${\displaystyle \mathbb {R} ^{n}}$, and let ${\displaystyle p\geq 1}$. A Young measure (with finite p-moments) is a family of Borel probability measures ${\displaystyle \{\nu _{x}:x\in U\}}$ on ${\displaystyle \mathbb {R} ^{m}}$ such that ${\displaystyle \int _{U}\int _{\mathbb {R} ^{m}}\|y\|^{p}d\nu _{x}(y)dx<+\infty }$.

Examples

Pointwise converging sequence

A trivial example of Young measure is when the sequence ${\displaystyle f_{n}}$ is bounded in ${\displaystyle L^{\infty }(U,\mathbb {R} ^{n})}$ and converges pointwise almost everywhere in ${\displaystyle U}$ to a function ${\displaystyle f}$. The Young measure is then the Dirac measure

${\displaystyle \nu _{x}=\delta _{f(x)},\quad x\in U.}$

Indeed, by dominated convergence theorem, ${\displaystyle F(f_{n}(x))}$ converges weakly* in ${\displaystyle L^{\infty }(U)}$ to

${\displaystyle F(f(x))=\int F(y)\,{\text{d}}\delta _{f(x)}}$

for any ${\displaystyle F\in C(\mathbb {R} ^{n})}$.

Sequence of sines

A less trivial example is a sequence

${\displaystyle f_{n}(x)=\sin(nx),\quad x\in (0,2\pi ).}$

The corresponding Young measure satisfies ^[4]

${\displaystyle \nu _{x}(E)={\frac {1}{\pi }}\int _{E\cap [-1,1]}{\frac {1}{\sqrt {1-y^{2}}}}\,{\text{d}}y,}$

for any measurable set ${\displaystyle E}$, independent of ${\displaystyle x\in (0,2\pi )}$. In other words, for any ${\displaystyle F\in C(\mathbb {R} ^{n})}$:

${\displaystyle F(f_{n}){\rightharpoonup }^{*}{\frac {1}{\pi }}\int _{-1}^{1}{\frac {F(y)}{\sqrt {1-y^{2}}}}\,{\text{d}}y}$

in ${\displaystyle L^{\infty }((0,2\pi ))}$. Here, the Young measure does not depend on ${\displaystyle x}$ and so the weak* limit is always a constant.

To see this intuitively, consider that at the limit of large ${\displaystyle n}$, a rectangle of ${\displaystyle [x,x+\delta x]\times [y,y+\delta y]}$ would capture a part of the curve of ${\displaystyle f_{n}}$. Take that captured part, and project it down to the x-axis. The length of that projection is ${\displaystyle {\frac {2\delta x\delta y}{\sqrt {1-y^{2}}}}}$, which means that ${\displaystyle \lim _{n}f_{n}}$ should look like a fine mist that has probability density ${\displaystyle {\frac {1}{\pi {\sqrt {1-y^{2}}}}}}$ at all ${\displaystyle x}$.

Minimizing sequence

For every asymptotically minimizing sequence ${\displaystyle u_{n}}$ of

${\displaystyle I(u)=\int _{0}^{1}(u'(x)^{2}-1)^{2}+u(x)^{2}dx}$

subject to ${\displaystyle u(0)=u(1)=0}$ (that is, the sequence satisfies ${\displaystyle \lim _{n\to +\infty }I(u_{n})=\inf _{u\in C^{1}([0,1])}I(u)}$), and perhaps after passing to a subsequence, the sequence of derivatives ${\displaystyle u'_{n}}$ generates Young measures of the form ${\displaystyle \nu _{x}={\frac {1}{2}}\delta _{-1}+{\frac {1}{2}}\delta _{1}}$. This captures the essential features of all minimizing sequences to this problem, namely, their derivatives ${\displaystyle u'_{k}(x)}$ will tend to concentrate along the minima ${\displaystyle \{-1,1\}}$ of the integrand ${\displaystyle (u'(x)^{2}-1)^{2}+u(x)^{2}}$.

If we take ${\displaystyle \lim _{n}{\frac {\sin(2\pi nt)}{2\pi n}}}$, then its limit has value zero, and derivative ${\displaystyle \nu (dy)={\frac {1}{\pi {\sqrt {1-y^{2}}}}}dy}$, which means ${\displaystyle \lim I={\frac {1}{\pi }}\int _{-1}^{+1}(1-y^{2})^{3/2}dy}$.

See also

• Convex compactification

References

1. Young, L. C. (1942). "Generalized Surfaces in the Calculus of Variations". Annals of Mathematics. 43 (1): 84–103. doi:10.2307/1968882. ISSN   0003-486X. JSTOR   1968882.
2. Balder, Erik J. "Lectures on Young measures." Cahiers de Mathématiques de la Décision 9517 (1995).
3. Pedregal, Pablo (1997). Parametrized measures and variational principles. Basel: Birkhäuser Verlag. ISBN   978-3-0348-8886-8. OCLC   812613013.
4. Dacorogna, Bernard (2006). Weak continuity and weak lower semicontinuity of non-linear functionals. Springer.

• Ball, J. M. (1989). "A version of the fundamental theorem for Young measures". In Rascle, M.; Serre, D.; Slemrod, M. (eds.). PDEs and Continuum Models of Phase Transitions : Proceedings of an NSF-CNRS Joint Seminar Held in Nice, France, January 18–22, 1988. Lecture Notes in Physics. Vol. 344. Berlin: Springer. pp. 207–215. doi:10.1007/BFb0024945. ISBN   3-540-51617-4. ISSN   0075-8450.
• Castaing, Charles; Raynaud de Fitte, Paul; Valadier, Michel (2004). Young Measures on Topological Spaces : With Applications in Control Theory and Probability Theory. Dordrecht: Kluwer Academic Publishers (now Springer). doi:10.1007/1-4020-1964-5. ISBN   978-1-4020-1963-0.
• L.C. Evans (1990). Weak convergence methods for nonlinear partial differential equations. Regional conference series in mathematics. American Mathematical Society.
• S. Müller (1999). Variational models for microstructure and phase transitions. Lecture Notes in Mathematics. Springer.
• P. Pedregal (1997). Parametrized Measures and Variational Principles. Basel: Birkhäuser. ISBN   978-3-0348-9815-7.
• T. Roubíček (2020). Relaxation in Optimization Theory and Variational Calculus (2nd ed.). Berlin: W. de Gruyter. doi:10.1515/9783110590852. ISBN   978-3-11-059085-2.
• Valadier, M. (1990). "Young measures". In Cellina, A. (ed.). Methods of Nonconvex Analysis. Lecture Notes in Mathematics. Vol. 1446. Berlin: Springer. pp. 152–188. doi:10.1007/BFb0084935. ISBN   3-540-53120-3.
• Young, L. C. (1937), "Generalized curves and the existence of an attained absolute minimum in the Calculus of Variations", Comptes Rendus des Séances de la Société des Sciences et des Lettres de Varsovie, Classe III, XXX (7–9): 211–234, JFM   63.1064.01, Zbl   0019.21901 , memoir presented by Stanisław Saks at the session of 16 December 1937 of the Warsaw Society of Sciences and Letters. The free PDF copy is made available by the RCIN –Digital Repository of the Scientifics Institutes.
• Young, L. C. (1969), Lectures on the Calculus of Variations and Optimal Control, Philadelphia–London–Toronto: W. B. Saunders, pp. xi+331, ISBN   9780721696409, MR   0259704, Zbl   0177.37801 .

External links

• "Young measure", Encyclopedia of Mathematics , EMS Press, 2001 [1994]