Time-dependent systems of generalized Young measures (Q2474746)

The authors extend the definition of Young measures to time-dependent systems of generalized Young measures. Here \((X, d)\) is a compact metric space, \(\lambda\) is a fixed Radon measure on \(X\) with supp( \(\lambda\) ) = \(X\), \(\Xi\) is a finite-dimensional Hilbert space, \(C^{hom}(X \times \Xi) = \{ f: X \times \Xi \to \mathbb{R} \; | f(x, \xi) \text{ positively homogeneous, of degree 1, in } \xi \}\) with norm \(\| f \|_{hom} = \sup \{ f(x, \xi): x \in X, \xi \text{ in the unit sphere of } \Xi \}\) and dual \(M_{*}(X \times \Xi)\), similar meanings to \(C^{hom}(X \times \Xi \times \mathbb{R})\) when \( \Xi\) is replaced by \( \Xi \times \mathbb{R}\) (here \(\eta\) will refer to the elements in \(\mathbb{R}\)). For a locally compact space \(E\), \(M_{b}(E)\) denotes the set of all bounded Radon measures on \(E\). \(< . >\) will be used for duality product. The set of Young measures is the set \(Y(X; \Xi) = \{ \nu \in M_{b}^{+}(X \times \Xi): \pi_{X}(\nu) = \lambda \} \) and the of generalized Young measures is the set \(GY(X; \Xi)= \{ \mu \in M_{*}^{+}(X \times \Xi \times \mathbb{R}): supp( \mu )\subset X \times \Xi \times [0, \infty ), \pi_{X}( \eta \mu) = \lambda \}\). For a time set \(A \subset \mathbb{R} \), a family of \( \mu = (\mu_{t_{1} \dots t_{m}})\) of generalized Young measures in \(GY(X; \Xi^{m})\), satisfying some compatibility conditions when \(t_{1}, \dots ,t_{m}\) run over finite sequences in \(A\) with \( t_{1} < \dots < t_{m}\), is called the compatible system of generalized Young measures and is denoted by \(SGY(A, X; \Xi)\). For a \( \mu \in SGY(A, X; \Xi)\), \( a, b \in A\), the variation of \(\mu\) is defined as \(Var(\mu; a, b)= \sup \{ \sum_{i=1}^{k} (<|\xi_{i} -\xi_{i-1} |, \mu_{t_{0} \dots t_{k}}(x, \xi_{0}, \dots , \xi_{k})> \}\), with \(a= t_{0} < \dots < t_{k}=b \) as the finite family \( t_{0}, \dots ,t_{k}\) varies in \(A\); if \(h: \Xi \to [0, \infty)\) is a positively one-homogeneous function, satisfying the triangle inequality, then \(Var_{h}(\mu; a, b)= \sup \{ \sum_{i=1}^{k} <h(\xi_{i} -\xi_{i-1}), \mu_{t_{0} \dots t_{k}}(x, \xi_{0}, \dots , \xi_{k})> \}\) with same conditions on \( t_{0}, \dots ,t_{k}\). For a \(T>0\), \( t_{0}, t \in [0, T]\) with \( t_{0} \neq t\), \(\mu \in SGY([0, T], X; \Xi)\) and a mapping \( q_{t_{0} t}: X \times \Xi \times \Xi \times \mathbb{R} \to X \times \Xi \times \mathbb{R}, \; q_{t_{0} t}(x, \xi_{1}, \xi_{2}, \eta)= (x, \frac{\xi_{2} -\xi_{1}}{t - t_{0}} ,\eta)\), the image measure \(q_{t_{0} t} (\mu_{t_{0} t})\) is in \(GY(X; \Xi)\); \( \mu\) is said to have \(weak^{*}\) derivative \(\mu_{t_{0}}^{.} \in GY(X; \Xi)\) at \(t_{0}\) if \(q_{t_{0} t} (\mu_{t_{0} t}) \to \mu_{t_{0}}^{.}\) \(weak^{*}\) in \(GY(X; \Xi)\) as \(t \to t_{0}\). Several results about these measures are proved. Two major results are: (i) Let \(T>0\), \( t_{0} \in [0, T]\), \( \{ \mu^{k} \}\) a sequence in \( SGY([0, T], X; \Xi)\) such that \(\sup_{k} Var(\mu^{k}; 0, T) \leq C, \; \sup_{k} \|\mu_{t_{0}}^{k} \|_{*} \leq C_{*}\) for some finite \(C, \; C_{*}\). Then there is a subsequence, again denoted by \( \{ \mu^{k} \}\), a \( \Theta \subset [0, T] \text{ with } 0 \in \Theta \text{ and } [0, T] \setminus \Theta\) at most countable, and a left continuous \( \mu \in SGY([0, T], X; \Xi)\) with \(Var(\mu; 0, T) \leq C\) and \(\| \mu_{t} \|_{*} \leq C + C_{*} \; \forall t \in [0, T]\) such that \(\mu_{t_{1} \dots t_{m}}^{k} \to \mu_{t_{1} \dots t_{m}} \) \( weak^{*} \) for all finite sequences \( t_{1}, \dots ,t_{m} \in \Theta\) with \(0 \leq t_{1} < \dots < t_{m} \leq T\). (ii) Let \(T>0\), \( \mu \in SGY([0, T], X; \Xi)\) satisfying \( Var(\mu; 0, T) < \infty\). Then the \(weak^{*}\) derivative \(\mu_{t}^{.}\) exists for \(a.e. \; t \in [0, T]\), \( t \to <f, \mu_{t}^{.}> \) is integrable on \([0, T]\) for every \( f \in C^{hom}(X \times \Xi \times \mathbb{R})\), and for any positively one-homogeneous function \(h: \Xi \to [0, \infty)\) satisfying the triangle inequality \(\int_{a}^{b} (h(\xi), \mu_{t}^{.}(x, \xi, \eta)) dt \leq Var_{h}(\mu; a, b)\) for \( 0 \leq a \leq b \leq T\).