Convergence of jump function with respect to the Skorokhod metric

The name of the pictureThe name of the pictureThe name of the pictureClash Royale CLAN TAG#URR8PPP











up vote
0
down vote

favorite












Consider the space $D := D([0, 1], mathbbN)$ of cadlag functions $f : [0, 1] to mathbbN$ equipped with the Skorokhod $J_1$-metric $d(f,g) := inf_lambda in Lambda max lVert lambda - id rVert, lVert f circ lambda - g rVert $ where $Lambda$ is the set of continuous strictly increasing functions $lambda : [0,1] to [0,1]$ with $lambda(0) = 0$ and $lambda(1) = 1$ and $lVert cdot rVert$ the supremum norm.
Let $x_n, x in D$ with $x_n to x$ in this metric. Denote by $tau_n$ and $tau$ the timepoint of the first jump of $x_n$ and $x$ respectively. Is it true that $tau_n to tau$ and $x_n(tau_n) to x(tau)$?










share|cite|improve this question



























    up vote
    0
    down vote

    favorite












    Consider the space $D := D([0, 1], mathbbN)$ of cadlag functions $f : [0, 1] to mathbbN$ equipped with the Skorokhod $J_1$-metric $d(f,g) := inf_lambda in Lambda max lVert lambda - id rVert, lVert f circ lambda - g rVert $ where $Lambda$ is the set of continuous strictly increasing functions $lambda : [0,1] to [0,1]$ with $lambda(0) = 0$ and $lambda(1) = 1$ and $lVert cdot rVert$ the supremum norm.
    Let $x_n, x in D$ with $x_n to x$ in this metric. Denote by $tau_n$ and $tau$ the timepoint of the first jump of $x_n$ and $x$ respectively. Is it true that $tau_n to tau$ and $x_n(tau_n) to x(tau)$?










    share|cite|improve this question

























      up vote
      0
      down vote

      favorite









      up vote
      0
      down vote

      favorite











      Consider the space $D := D([0, 1], mathbbN)$ of cadlag functions $f : [0, 1] to mathbbN$ equipped with the Skorokhod $J_1$-metric $d(f,g) := inf_lambda in Lambda max lVert lambda - id rVert, lVert f circ lambda - g rVert $ where $Lambda$ is the set of continuous strictly increasing functions $lambda : [0,1] to [0,1]$ with $lambda(0) = 0$ and $lambda(1) = 1$ and $lVert cdot rVert$ the supremum norm.
      Let $x_n, x in D$ with $x_n to x$ in this metric. Denote by $tau_n$ and $tau$ the timepoint of the first jump of $x_n$ and $x$ respectively. Is it true that $tau_n to tau$ and $x_n(tau_n) to x(tau)$?










      share|cite|improve this question















      Consider the space $D := D([0, 1], mathbbN)$ of cadlag functions $f : [0, 1] to mathbbN$ equipped with the Skorokhod $J_1$-metric $d(f,g) := inf_lambda in Lambda max lVert lambda - id rVert, lVert f circ lambda - g rVert $ where $Lambda$ is the set of continuous strictly increasing functions $lambda : [0,1] to [0,1]$ with $lambda(0) = 0$ and $lambda(1) = 1$ and $lVert cdot rVert$ the supremum norm.
      Let $x_n, x in D$ with $x_n to x$ in this metric. Denote by $tau_n$ and $tau$ the timepoint of the first jump of $x_n$ and $x$ respectively. Is it true that $tau_n to tau$ and $x_n(tau_n) to x(tau)$?







      general-topology skorohod-space






      share|cite|improve this question















      share|cite|improve this question













      share|cite|improve this question




      share|cite|improve this question








      edited Sep 4 at 15:11









      saz

      73.9k553113




      73.9k553113










      asked Sep 3 at 13:48









      yadaddy

      1,231816




      1,231816




















          1 Answer
          1






          active

          oldest

          votes

















          up vote
          2
          down vote



          accepted











          Lemma: If $x_n to x$ in $J_1$ topology and $x$ is not constant, then $tau_n to tau$ and $x_n(tau_n) to x(tau)$.




          Proof: Since $x_n to x$ in $J_1$ we can find a sequence $(lambda_n)_n in mathbbN$ of continuous strictly increasing functions $lambda_n: [0,1] to [0,1]$ with $lambda_n(0)=0$, $lambda_n(1)=1$ such that $$|lambda_n-textid|_infty to 0 qquad |x_n circ lambda_n-x|_infty to 0.$$ In particular, there exists $N in mathbbN$ such that $$|x_n circ lambda_n-x|_infty < 1 quad textfor all $n geq N$,$$ and since the functions are taking values in $mathbbN$, this already implies $$ forall n geq N , forall t in [0,1]: quad x_n(lambda_n(t))= x(t). tag1$$ As $x$ is right-continuous, non-constant and takes values in $mathbbN$, we know that the first jump time $tau$ of $x$ satisfies $$tau = inft in (0,1]: x(0) neq x(t) in (0,1].$$ It follows from $(1)$ that $$tau_n = lambda_n(tau)$$ and $$x_n(tau_n) = x_n(lambda_n(tau)) = x(tau)$$ for all $n geq N$. In particular, $x_n(lambda_n(tau)) to x(tau)$ and $$tau_n =lambda_n(tau) xrightarrown to infty tau.$$






          share|cite|improve this answer






















          • Please observe that the range space of the cadlag functions is discrete and countable.
            – yadaddy
            Sep 4 at 6:11










          • @yadaddy Sorry, I missed that; see my edited answer.
            – saz
            Sep 4 at 15:10










          • Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
            – yadaddy
            Sep 5 at 7:41







          • 1




            @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
            – saz
            Sep 5 at 7:52











          • Yes, that helped.
            – yadaddy
            Sep 5 at 8:13










          Your Answer




          StackExchange.ifUsing("editor", function ()
          return StackExchange.using("mathjaxEditing", function ()
          StackExchange.MarkdownEditor.creationCallbacks.add(function (editor, postfix)
          StackExchange.mathjaxEditing.prepareWmdForMathJax(editor, postfix, [["$", "$"], ["\\(","\\)"]]);
          );
          );
          , "mathjax-editing");

          StackExchange.ready(function()
          var channelOptions =
          tags: "".split(" "),
          id: "69"
          ;
          initTagRenderer("".split(" "), "".split(" "), channelOptions);

          StackExchange.using("externalEditor", function()
          // Have to fire editor after snippets, if snippets enabled
          if (StackExchange.settings.snippets.snippetsEnabled)
          StackExchange.using("snippets", function()
          createEditor();
          );

          else
          createEditor();

          );

          function createEditor()
          StackExchange.prepareEditor(
          heartbeatType: 'answer',
          convertImagesToLinks: true,
          noModals: false,
          showLowRepImageUploadWarning: true,
          reputationToPostImages: 10,
          bindNavPrevention: true,
          postfix: "",
          noCode: true, onDemand: true,
          discardSelector: ".discard-answer"
          ,immediatelyShowMarkdownHelp:true
          );



          );













           

          draft saved


          draft discarded


















          StackExchange.ready(
          function ()
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2903872%2fconvergence-of-jump-function-with-respect-to-the-skorokhod-metric%23new-answer', 'question_page');

          );

          Post as a guest






























          1 Answer
          1






          active

          oldest

          votes








          1 Answer
          1






          active

          oldest

          votes









          active

          oldest

          votes






          active

          oldest

          votes








          up vote
          2
          down vote



          accepted











          Lemma: If $x_n to x$ in $J_1$ topology and $x$ is not constant, then $tau_n to tau$ and $x_n(tau_n) to x(tau)$.




          Proof: Since $x_n to x$ in $J_1$ we can find a sequence $(lambda_n)_n in mathbbN$ of continuous strictly increasing functions $lambda_n: [0,1] to [0,1]$ with $lambda_n(0)=0$, $lambda_n(1)=1$ such that $$|lambda_n-textid|_infty to 0 qquad |x_n circ lambda_n-x|_infty to 0.$$ In particular, there exists $N in mathbbN$ such that $$|x_n circ lambda_n-x|_infty < 1 quad textfor all $n geq N$,$$ and since the functions are taking values in $mathbbN$, this already implies $$ forall n geq N , forall t in [0,1]: quad x_n(lambda_n(t))= x(t). tag1$$ As $x$ is right-continuous, non-constant and takes values in $mathbbN$, we know that the first jump time $tau$ of $x$ satisfies $$tau = inft in (0,1]: x(0) neq x(t) in (0,1].$$ It follows from $(1)$ that $$tau_n = lambda_n(tau)$$ and $$x_n(tau_n) = x_n(lambda_n(tau)) = x(tau)$$ for all $n geq N$. In particular, $x_n(lambda_n(tau)) to x(tau)$ and $$tau_n =lambda_n(tau) xrightarrown to infty tau.$$






          share|cite|improve this answer






















          • Please observe that the range space of the cadlag functions is discrete and countable.
            – yadaddy
            Sep 4 at 6:11










          • @yadaddy Sorry, I missed that; see my edited answer.
            – saz
            Sep 4 at 15:10










          • Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
            – yadaddy
            Sep 5 at 7:41







          • 1




            @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
            – saz
            Sep 5 at 7:52











          • Yes, that helped.
            – yadaddy
            Sep 5 at 8:13














          up vote
          2
          down vote



          accepted











          Lemma: If $x_n to x$ in $J_1$ topology and $x$ is not constant, then $tau_n to tau$ and $x_n(tau_n) to x(tau)$.




          Proof: Since $x_n to x$ in $J_1$ we can find a sequence $(lambda_n)_n in mathbbN$ of continuous strictly increasing functions $lambda_n: [0,1] to [0,1]$ with $lambda_n(0)=0$, $lambda_n(1)=1$ such that $$|lambda_n-textid|_infty to 0 qquad |x_n circ lambda_n-x|_infty to 0.$$ In particular, there exists $N in mathbbN$ such that $$|x_n circ lambda_n-x|_infty < 1 quad textfor all $n geq N$,$$ and since the functions are taking values in $mathbbN$, this already implies $$ forall n geq N , forall t in [0,1]: quad x_n(lambda_n(t))= x(t). tag1$$ As $x$ is right-continuous, non-constant and takes values in $mathbbN$, we know that the first jump time $tau$ of $x$ satisfies $$tau = inft in (0,1]: x(0) neq x(t) in (0,1].$$ It follows from $(1)$ that $$tau_n = lambda_n(tau)$$ and $$x_n(tau_n) = x_n(lambda_n(tau)) = x(tau)$$ for all $n geq N$. In particular, $x_n(lambda_n(tau)) to x(tau)$ and $$tau_n =lambda_n(tau) xrightarrown to infty tau.$$






          share|cite|improve this answer






















          • Please observe that the range space of the cadlag functions is discrete and countable.
            – yadaddy
            Sep 4 at 6:11










          • @yadaddy Sorry, I missed that; see my edited answer.
            – saz
            Sep 4 at 15:10










          • Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
            – yadaddy
            Sep 5 at 7:41







          • 1




            @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
            – saz
            Sep 5 at 7:52











          • Yes, that helped.
            – yadaddy
            Sep 5 at 8:13












          up vote
          2
          down vote



          accepted







          up vote
          2
          down vote



          accepted







          Lemma: If $x_n to x$ in $J_1$ topology and $x$ is not constant, then $tau_n to tau$ and $x_n(tau_n) to x(tau)$.




          Proof: Since $x_n to x$ in $J_1$ we can find a sequence $(lambda_n)_n in mathbbN$ of continuous strictly increasing functions $lambda_n: [0,1] to [0,1]$ with $lambda_n(0)=0$, $lambda_n(1)=1$ such that $$|lambda_n-textid|_infty to 0 qquad |x_n circ lambda_n-x|_infty to 0.$$ In particular, there exists $N in mathbbN$ such that $$|x_n circ lambda_n-x|_infty < 1 quad textfor all $n geq N$,$$ and since the functions are taking values in $mathbbN$, this already implies $$ forall n geq N , forall t in [0,1]: quad x_n(lambda_n(t))= x(t). tag1$$ As $x$ is right-continuous, non-constant and takes values in $mathbbN$, we know that the first jump time $tau$ of $x$ satisfies $$tau = inft in (0,1]: x(0) neq x(t) in (0,1].$$ It follows from $(1)$ that $$tau_n = lambda_n(tau)$$ and $$x_n(tau_n) = x_n(lambda_n(tau)) = x(tau)$$ for all $n geq N$. In particular, $x_n(lambda_n(tau)) to x(tau)$ and $$tau_n =lambda_n(tau) xrightarrown to infty tau.$$






          share|cite|improve this answer















          Lemma: If $x_n to x$ in $J_1$ topology and $x$ is not constant, then $tau_n to tau$ and $x_n(tau_n) to x(tau)$.




          Proof: Since $x_n to x$ in $J_1$ we can find a sequence $(lambda_n)_n in mathbbN$ of continuous strictly increasing functions $lambda_n: [0,1] to [0,1]$ with $lambda_n(0)=0$, $lambda_n(1)=1$ such that $$|lambda_n-textid|_infty to 0 qquad |x_n circ lambda_n-x|_infty to 0.$$ In particular, there exists $N in mathbbN$ such that $$|x_n circ lambda_n-x|_infty < 1 quad textfor all $n geq N$,$$ and since the functions are taking values in $mathbbN$, this already implies $$ forall n geq N , forall t in [0,1]: quad x_n(lambda_n(t))= x(t). tag1$$ As $x$ is right-continuous, non-constant and takes values in $mathbbN$, we know that the first jump time $tau$ of $x$ satisfies $$tau = inft in (0,1]: x(0) neq x(t) in (0,1].$$ It follows from $(1)$ that $$tau_n = lambda_n(tau)$$ and $$x_n(tau_n) = x_n(lambda_n(tau)) = x(tau)$$ for all $n geq N$. In particular, $x_n(lambda_n(tau)) to x(tau)$ and $$tau_n =lambda_n(tau) xrightarrown to infty tau.$$







          share|cite|improve this answer














          share|cite|improve this answer



          share|cite|improve this answer








          edited Sep 4 at 15:10

























          answered Sep 4 at 6:08









          saz

          73.9k553113




          73.9k553113











          • Please observe that the range space of the cadlag functions is discrete and countable.
            – yadaddy
            Sep 4 at 6:11










          • @yadaddy Sorry, I missed that; see my edited answer.
            – saz
            Sep 4 at 15:10










          • Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
            – yadaddy
            Sep 5 at 7:41







          • 1




            @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
            – saz
            Sep 5 at 7:52











          • Yes, that helped.
            – yadaddy
            Sep 5 at 8:13
















          • Please observe that the range space of the cadlag functions is discrete and countable.
            – yadaddy
            Sep 4 at 6:11










          • @yadaddy Sorry, I missed that; see my edited answer.
            – saz
            Sep 4 at 15:10










          • Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
            – yadaddy
            Sep 5 at 7:41







          • 1




            @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
            – saz
            Sep 5 at 7:52











          • Yes, that helped.
            – yadaddy
            Sep 5 at 8:13















          Please observe that the range space of the cadlag functions is discrete and countable.
          – yadaddy
          Sep 4 at 6:11




          Please observe that the range space of the cadlag functions is discrete and countable.
          – yadaddy
          Sep 4 at 6:11












          @yadaddy Sorry, I missed that; see my edited answer.
          – saz
          Sep 4 at 15:10




          @yadaddy Sorry, I missed that; see my edited answer.
          – saz
          Sep 4 at 15:10












          Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
          – yadaddy
          Sep 5 at 7:41





          Thank you. I see that $x_n(lambda_n(tau)) = x(tau)$ for all $n geq N$, but I don't see why $tau_n = lambda_n(tau)$. The time shifts $lambda_n$ that are given to us by $J_1$-convergence need not be those that map the jump time point $tau$ of $x$ to the jump timepoints $tau_n$ of $x_n$.
          – yadaddy
          Sep 5 at 7:41





          1




          1




          @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
          – saz
          Sep 5 at 7:52





          @yadaddy For $s in [0,tau)$, we have $x(s)=x(0)$, and it follows from $(1)$ that $x_n(s)=x(0)$ for any $s in [0,lambda_n(tau))$. At $s=lambda_n(tau)$, it follows again from (1) that $x_n$ jumps and the new position is exactly $x(tau)$. If this makes things easier for you, you may note that $lambda_n$ is strictly increasing (hence invertible), and then $(1)$ becomes $x_n(s) = x(lambda_n^-1(s))$. By looking at the "trajectory" of $x$ you can recover all the information about $x_n$.
          – saz
          Sep 5 at 7:52













          Yes, that helped.
          – yadaddy
          Sep 5 at 8:13




          Yes, that helped.
          – yadaddy
          Sep 5 at 8:13

















           

          draft saved


          draft discarded















































           


          draft saved


          draft discarded














          StackExchange.ready(
          function ()
          StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f2903872%2fconvergence-of-jump-function-with-respect-to-the-skorokhod-metric%23new-answer', 'question_page');

          );

          Post as a guest













































































          這個網誌中的熱門文章

          How to combine Bézier curves to a surface?

          Mutual Information Always Non-negative

          Why am i infinitely getting the same tweet with the Twitter Search API?