Verifying uniform convergence for a convolution












0












$begingroup$


In class my professor claimed that that if $f: mathbb{R} rightarrow mathbb{R}$ is a continuous function on a compact set, then the convolution of $j_{epsilon}(x)$ and $f$ converges uniformly to $f$ as $epsilon rightarrow 0$, where $j_{epsilon}(x)$ is a smooth function with compact support that integrates to 1 over $mathbb{R}^n$.



I would like to verify this for $f = exp({-x^2})chi_{[-1,1]}$ and $j_{epsilon}(x) = $ $chi_{[-1,1]}(x)exp({-x^2 over epsilon^2}) over sqrt{2piepsilon}$, but I have no idea where to start. Any help is appreciated.



Edit: I accidentally left out the characteristic function on $j_{epsilon}(x)$. I've added it now.










share|cite|improve this question











$endgroup$








  • 1




    $begingroup$
    You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:36










  • $begingroup$
    What kind of condition?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:38










  • $begingroup$
    How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:43










  • $begingroup$
    Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:46
















0












$begingroup$


In class my professor claimed that that if $f: mathbb{R} rightarrow mathbb{R}$ is a continuous function on a compact set, then the convolution of $j_{epsilon}(x)$ and $f$ converges uniformly to $f$ as $epsilon rightarrow 0$, where $j_{epsilon}(x)$ is a smooth function with compact support that integrates to 1 over $mathbb{R}^n$.



I would like to verify this for $f = exp({-x^2})chi_{[-1,1]}$ and $j_{epsilon}(x) = $ $chi_{[-1,1]}(x)exp({-x^2 over epsilon^2}) over sqrt{2piepsilon}$, but I have no idea where to start. Any help is appreciated.



Edit: I accidentally left out the characteristic function on $j_{epsilon}(x)$. I've added it now.










share|cite|improve this question











$endgroup$








  • 1




    $begingroup$
    You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:36










  • $begingroup$
    What kind of condition?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:38










  • $begingroup$
    How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:43










  • $begingroup$
    Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:46














0












0








0


1



$begingroup$


In class my professor claimed that that if $f: mathbb{R} rightarrow mathbb{R}$ is a continuous function on a compact set, then the convolution of $j_{epsilon}(x)$ and $f$ converges uniformly to $f$ as $epsilon rightarrow 0$, where $j_{epsilon}(x)$ is a smooth function with compact support that integrates to 1 over $mathbb{R}^n$.



I would like to verify this for $f = exp({-x^2})chi_{[-1,1]}$ and $j_{epsilon}(x) = $ $chi_{[-1,1]}(x)exp({-x^2 over epsilon^2}) over sqrt{2piepsilon}$, but I have no idea where to start. Any help is appreciated.



Edit: I accidentally left out the characteristic function on $j_{epsilon}(x)$. I've added it now.










share|cite|improve this question











$endgroup$




In class my professor claimed that that if $f: mathbb{R} rightarrow mathbb{R}$ is a continuous function on a compact set, then the convolution of $j_{epsilon}(x)$ and $f$ converges uniformly to $f$ as $epsilon rightarrow 0$, where $j_{epsilon}(x)$ is a smooth function with compact support that integrates to 1 over $mathbb{R}^n$.



I would like to verify this for $f = exp({-x^2})chi_{[-1,1]}$ and $j_{epsilon}(x) = $ $chi_{[-1,1]}(x)exp({-x^2 over epsilon^2}) over sqrt{2piepsilon}$, but I have no idea where to start. Any help is appreciated.



Edit: I accidentally left out the characteristic function on $j_{epsilon}(x)$. I've added it now.







real-analysis convolution






share|cite|improve this question















share|cite|improve this question













share|cite|improve this question




share|cite|improve this question








edited Dec 19 '18 at 10:52







Taliant

















asked Dec 19 '18 at 10:29









TaliantTaliant

839




839








  • 1




    $begingroup$
    You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:36










  • $begingroup$
    What kind of condition?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:38










  • $begingroup$
    How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:43










  • $begingroup$
    Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:46














  • 1




    $begingroup$
    You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:36










  • $begingroup$
    What kind of condition?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:38










  • $begingroup$
    How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
    $endgroup$
    – Mees de Vries
    Dec 19 '18 at 10:43










  • $begingroup$
    Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
    $endgroup$
    – Taliant
    Dec 19 '18 at 10:46








1




1




$begingroup$
You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
$endgroup$
– Mees de Vries
Dec 19 '18 at 10:36




$begingroup$
You will need at least one more condition on $j_epsilon$ as the question stands, because right now there is no condition on $j_epsilon$ that depends on $epsilon$.
$endgroup$
– Mees de Vries
Dec 19 '18 at 10:36












$begingroup$
What kind of condition?
$endgroup$
– Taliant
Dec 19 '18 at 10:38




$begingroup$
What kind of condition?
$endgroup$
– Taliant
Dec 19 '18 at 10:38












$begingroup$
How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
$endgroup$
– Mees de Vries
Dec 19 '18 at 10:43




$begingroup$
How would I know? It's not my question. One that would probably suffice is to have the support of $j$ within $[-epsilon, epsilon]$. (However, note that your $j_epsilon$ do not even have compact support....)
$endgroup$
– Mees de Vries
Dec 19 '18 at 10:43












$begingroup$
Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
$endgroup$
– Taliant
Dec 19 '18 at 10:46




$begingroup$
Let's suppose that it does (I'll edit the question to reflect that). Can I show uniform convergence in that case?
$endgroup$
– Taliant
Dec 19 '18 at 10:46










1 Answer
1






active

oldest

votes


















0












$begingroup$

Note that since support of $f$ is compact, $f$ is uniformly continuous and bounded. Let $j(x)$ be an arbitrary function such that
$$
int_mathbb{R} |j(x)|dx=lVert jrVert_1 <infty, ;int_mathbb{R} j(x)dx=1.
$$
Let us define $j_epsilon(x) = frac{1}{epsilon}j(frac{x}{epsilon})$ as a dilation of $j$. Then, it holds that
$$
(f*j_epsilon)(x) = int_mathbb{R} f(x-y)j_epsilon(y)dyto f(x)
$$
uniformly as $epsilon to 0$. To see this, observe that for any fixed $delta>0$, it holds
$$begin{eqnarray}
|(f*j_epsilon)(x)-f(x)|&= & |int_mathbb{R} (f(x-y)-f(x))j_epsilon(y)dy|\
&leq&int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dy+ int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dy.
end{eqnarray}$$
The first term can be estimated as
$$
int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dyleq lVert jrVert_1cdotsup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|.
$$
The second term can be estimated as
$$
int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dyleq 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$
From this, by taking supremum on the LHS over all $x$, we have
$$
lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)| + 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$

As $epsilon to 0$, we have $frac{delta}{epsilon}to infty$, and hence that
$$
int_{|z|>frac{delta}{epsilon}}|j(z)|dz to 0,
$$
by dominated convergence theorem. This implies
$$
limsup_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|,
$$
for all $delta>0$. Finally, by taking $delta to 0$, we get
$$
lim_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_infty =0,
$$
as desired.



$textbf{EDIT:}$ As a corollary, $$j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$$ is suitable for your purpose. You don't need to multiply $chi_{[-1,1]}(x)$, though it doesn't matter for your purpose.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:32










  • $begingroup$
    Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
    $endgroup$
    – Song
    Dec 19 '18 at 11:33












  • $begingroup$
    Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:39










  • $begingroup$
    Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
    $endgroup$
    – Song
    Dec 19 '18 at 11:41












  • $begingroup$
    Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:56











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',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
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%2f3046245%2fverifying-uniform-convergence-for-a-convolution%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























1 Answer
1






active

oldest

votes








1 Answer
1






active

oldest

votes









active

oldest

votes






active

oldest

votes









0












$begingroup$

Note that since support of $f$ is compact, $f$ is uniformly continuous and bounded. Let $j(x)$ be an arbitrary function such that
$$
int_mathbb{R} |j(x)|dx=lVert jrVert_1 <infty, ;int_mathbb{R} j(x)dx=1.
$$
Let us define $j_epsilon(x) = frac{1}{epsilon}j(frac{x}{epsilon})$ as a dilation of $j$. Then, it holds that
$$
(f*j_epsilon)(x) = int_mathbb{R} f(x-y)j_epsilon(y)dyto f(x)
$$
uniformly as $epsilon to 0$. To see this, observe that for any fixed $delta>0$, it holds
$$begin{eqnarray}
|(f*j_epsilon)(x)-f(x)|&= & |int_mathbb{R} (f(x-y)-f(x))j_epsilon(y)dy|\
&leq&int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dy+ int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dy.
end{eqnarray}$$
The first term can be estimated as
$$
int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dyleq lVert jrVert_1cdotsup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|.
$$
The second term can be estimated as
$$
int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dyleq 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$
From this, by taking supremum on the LHS over all $x$, we have
$$
lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)| + 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$

As $epsilon to 0$, we have $frac{delta}{epsilon}to infty$, and hence that
$$
int_{|z|>frac{delta}{epsilon}}|j(z)|dz to 0,
$$
by dominated convergence theorem. This implies
$$
limsup_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|,
$$
for all $delta>0$. Finally, by taking $delta to 0$, we get
$$
lim_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_infty =0,
$$
as desired.



$textbf{EDIT:}$ As a corollary, $$j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$$ is suitable for your purpose. You don't need to multiply $chi_{[-1,1]}(x)$, though it doesn't matter for your purpose.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:32










  • $begingroup$
    Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
    $endgroup$
    – Song
    Dec 19 '18 at 11:33












  • $begingroup$
    Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:39










  • $begingroup$
    Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
    $endgroup$
    – Song
    Dec 19 '18 at 11:41












  • $begingroup$
    Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:56
















0












$begingroup$

Note that since support of $f$ is compact, $f$ is uniformly continuous and bounded. Let $j(x)$ be an arbitrary function such that
$$
int_mathbb{R} |j(x)|dx=lVert jrVert_1 <infty, ;int_mathbb{R} j(x)dx=1.
$$
Let us define $j_epsilon(x) = frac{1}{epsilon}j(frac{x}{epsilon})$ as a dilation of $j$. Then, it holds that
$$
(f*j_epsilon)(x) = int_mathbb{R} f(x-y)j_epsilon(y)dyto f(x)
$$
uniformly as $epsilon to 0$. To see this, observe that for any fixed $delta>0$, it holds
$$begin{eqnarray}
|(f*j_epsilon)(x)-f(x)|&= & |int_mathbb{R} (f(x-y)-f(x))j_epsilon(y)dy|\
&leq&int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dy+ int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dy.
end{eqnarray}$$
The first term can be estimated as
$$
int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dyleq lVert jrVert_1cdotsup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|.
$$
The second term can be estimated as
$$
int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dyleq 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$
From this, by taking supremum on the LHS over all $x$, we have
$$
lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)| + 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$

As $epsilon to 0$, we have $frac{delta}{epsilon}to infty$, and hence that
$$
int_{|z|>frac{delta}{epsilon}}|j(z)|dz to 0,
$$
by dominated convergence theorem. This implies
$$
limsup_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|,
$$
for all $delta>0$. Finally, by taking $delta to 0$, we get
$$
lim_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_infty =0,
$$
as desired.



$textbf{EDIT:}$ As a corollary, $$j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$$ is suitable for your purpose. You don't need to multiply $chi_{[-1,1]}(x)$, though it doesn't matter for your purpose.






share|cite|improve this answer











$endgroup$













  • $begingroup$
    Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:32










  • $begingroup$
    Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
    $endgroup$
    – Song
    Dec 19 '18 at 11:33












  • $begingroup$
    Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:39










  • $begingroup$
    Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
    $endgroup$
    – Song
    Dec 19 '18 at 11:41












  • $begingroup$
    Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:56














0












0








0





$begingroup$

Note that since support of $f$ is compact, $f$ is uniformly continuous and bounded. Let $j(x)$ be an arbitrary function such that
$$
int_mathbb{R} |j(x)|dx=lVert jrVert_1 <infty, ;int_mathbb{R} j(x)dx=1.
$$
Let us define $j_epsilon(x) = frac{1}{epsilon}j(frac{x}{epsilon})$ as a dilation of $j$. Then, it holds that
$$
(f*j_epsilon)(x) = int_mathbb{R} f(x-y)j_epsilon(y)dyto f(x)
$$
uniformly as $epsilon to 0$. To see this, observe that for any fixed $delta>0$, it holds
$$begin{eqnarray}
|(f*j_epsilon)(x)-f(x)|&= & |int_mathbb{R} (f(x-y)-f(x))j_epsilon(y)dy|\
&leq&int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dy+ int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dy.
end{eqnarray}$$
The first term can be estimated as
$$
int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dyleq lVert jrVert_1cdotsup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|.
$$
The second term can be estimated as
$$
int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dyleq 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$
From this, by taking supremum on the LHS over all $x$, we have
$$
lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)| + 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$

As $epsilon to 0$, we have $frac{delta}{epsilon}to infty$, and hence that
$$
int_{|z|>frac{delta}{epsilon}}|j(z)|dz to 0,
$$
by dominated convergence theorem. This implies
$$
limsup_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|,
$$
for all $delta>0$. Finally, by taking $delta to 0$, we get
$$
lim_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_infty =0,
$$
as desired.



$textbf{EDIT:}$ As a corollary, $$j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$$ is suitable for your purpose. You don't need to multiply $chi_{[-1,1]}(x)$, though it doesn't matter for your purpose.






share|cite|improve this answer











$endgroup$



Note that since support of $f$ is compact, $f$ is uniformly continuous and bounded. Let $j(x)$ be an arbitrary function such that
$$
int_mathbb{R} |j(x)|dx=lVert jrVert_1 <infty, ;int_mathbb{R} j(x)dx=1.
$$
Let us define $j_epsilon(x) = frac{1}{epsilon}j(frac{x}{epsilon})$ as a dilation of $j$. Then, it holds that
$$
(f*j_epsilon)(x) = int_mathbb{R} f(x-y)j_epsilon(y)dyto f(x)
$$
uniformly as $epsilon to 0$. To see this, observe that for any fixed $delta>0$, it holds
$$begin{eqnarray}
|(f*j_epsilon)(x)-f(x)|&= & |int_mathbb{R} (f(x-y)-f(x))j_epsilon(y)dy|\
&leq&int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dy+ int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dy.
end{eqnarray}$$
The first term can be estimated as
$$
int_{|y|<delta} |f(x-y)-f(x)||j_epsilon(y)|dyleq lVert jrVert_1cdotsup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|.
$$
The second term can be estimated as
$$
int_{|y|geqdelta} |f(x-y)-f(x)||j_epsilon(y)|dyleq 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$
From this, by taking supremum on the LHS over all $x$, we have
$$
lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)| + 2sup_{xinmathbb{R}}|f(x)|cdotint_{|z|>frac{delta}{epsilon}}|j(z)|dz.
$$

As $epsilon to 0$, we have $frac{delta}{epsilon}to infty$, and hence that
$$
int_{|z|>frac{delta}{epsilon}}|j(z)|dz to 0,
$$
by dominated convergence theorem. This implies
$$
limsup_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_inftyleq lVert jrVert_1cdot sup_{xinmathbb{R},|y|<delta}|f(x)-f(x-y)|,
$$
for all $delta>0$. Finally, by taking $delta to 0$, we get
$$
lim_{epsilonto 0}lVert(f*j_epsilon)(cdot)-f(cdot)rVert_infty =0,
$$
as desired.



$textbf{EDIT:}$ As a corollary, $$j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$$ is suitable for your purpose. You don't need to multiply $chi_{[-1,1]}(x)$, though it doesn't matter for your purpose.







share|cite|improve this answer














share|cite|improve this answer



share|cite|improve this answer








edited Dec 19 '18 at 11:27

























answered Dec 19 '18 at 11:13









SongSong

16.2k1739




16.2k1739












  • $begingroup$
    Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:32










  • $begingroup$
    Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
    $endgroup$
    – Song
    Dec 19 '18 at 11:33












  • $begingroup$
    Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:39










  • $begingroup$
    Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
    $endgroup$
    – Song
    Dec 19 '18 at 11:41












  • $begingroup$
    Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:56


















  • $begingroup$
    Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:32










  • $begingroup$
    Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
    $endgroup$
    – Song
    Dec 19 '18 at 11:33












  • $begingroup$
    Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:39










  • $begingroup$
    Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
    $endgroup$
    – Song
    Dec 19 '18 at 11:41












  • $begingroup$
    Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
    $endgroup$
    – Taliant
    Dec 19 '18 at 11:56
















$begingroup$
Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
$endgroup$
– Taliant
Dec 19 '18 at 11:32




$begingroup$
Please excuse this very basic question, but how is $j_epsilon(x) = frac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}$ compactly supported? Or is having a compact support not necessary here?
$endgroup$
– Taliant
Dec 19 '18 at 11:32












$begingroup$
Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
$endgroup$
– Song
Dec 19 '18 at 11:33






$begingroup$
Yes, $j_epsilon$ does not have to be compactly supported. But if you want, you can truncate it as $cfrac{1}{sqrt{2pi}epsilon}e^{-frac{x^2}{2epsilon^2}}chi_{[-1,1]}(frac{x}{epsilon})$. The constant $c$ is multiplied to make $int j_epsilon(x)dx = 1$.
$endgroup$
– Song
Dec 19 '18 at 11:33














$begingroup$
Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
$endgroup$
– Taliant
Dec 19 '18 at 11:39




$begingroup$
Another question: how do you do the estimation of the first and second term? It looks like Holder's inequality, but I can't tell.
$endgroup$
– Taliant
Dec 19 '18 at 11:39












$begingroup$
Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
$endgroup$
– Song
Dec 19 '18 at 11:41






$begingroup$
Both estimates use the same principle: if $phi$ is such that $|phi|leq M$ and $psi$ is such that $int |psi|dx = K$, then $|int phicdotpsi dx|leq int |phi|cdot|psi| dxleq MK.$
$endgroup$
– Song
Dec 19 '18 at 11:41














$begingroup$
Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
$endgroup$
– Taliant
Dec 19 '18 at 11:56




$begingroup$
Okay, that makes sense. Two other questions: how is the variable $z$ defined, and how is the limsup inequality justified?
$endgroup$
– Taliant
Dec 19 '18 at 11:56


















draft saved

draft discarded




















































Thanks for contributing an answer to Mathematics Stack Exchange!


  • Please be sure to answer the question. Provide details and share your research!

But avoid



  • Asking for help, clarification, or responding to other answers.

  • Making statements based on opinion; back them up with references or personal experience.


Use MathJax to format equations. MathJax reference.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3046245%2fverifying-uniform-convergence-for-a-convolution%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown





















































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown

































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown







Popular posts from this blog

Probability when a professor distributes a quiz and homework assignment to a class of n students.

Aardman Animations

Are they similar matrix