Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$...
up vote
0
down vote
favorite
GIVEN
Let $K subset mathbb{R}^n$ be a nonempty, closed and convex set. The associated distance function is $d_K$. Find the subdifferential $partial d_K(x_0)$ for all $x_0 in K$.
USEFUL DEFINITIONS
Convex Subdifferential
The subdifferential of a proper, lower semi-continuous, and convex function $f$ at a point $x_0 in text{dom}(f)$ is:
$$
partial f (x_0) = big{ z in mathbb{R}^n : f(x) geq f(x_0) + langle z, x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Normal Cone
The normal cone of a nonempty, closed, and convex set $K$ at a point $x_0 in K$ is:
$$
N_K(x_0)=big{ z:langle z, ;x-x_0 rangle leq 0,; forall x in Kbig}
$$
CONTEXT
After some research I have found out that this is a well-known result: $partial d_K (x_0) = bar{B}cap N_K (x_0)$ for all $x_0 in K$. Here $bar{B}$ refers to the closed unit ball.
Note that this is true for Banach spaces. I am unsure of the result in my case. I also cannot use infimal convolution as I have not learned it yet.
However, I could not find any proof of it anywhere, and thus I am trying to prove that $Vert z Vert leq 1$ (giving $z in bar{B}$) and that $langle z, ; x - x_0 rangle leq 0$ (giving $z in N_K(x_0)$). Here $z in partial d_K (x_0)$.
ATTEMPT
Let $x_0$ be a fixed point in $K$.
The distance function $d_K(x) = inf_{y in K} Vert x - yVert$ over a convex set $K$ is convex, proper and lower semi-continuous. Since $x_0 in K$ then $d_K (x_0) = 0$.
Using the above definition, obtain:
$$
partial d_K (x_0) = big{ zeta in mathbb{R}^n : d_K(x) geq langle z, ; x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Part 1 — Normal Cone
Case 1: $x in K$
Here, $d_K (x) = 0$ and the above equation gives $partial d_K(x_0)=N_K (x_0)$.
Case 2: $x notin K$
Here, $d_K (x) = Vert x - pVert > 0$, where $p$ is the projection of $x$ onto $K$.
I am completely stuck here and have tried multiple failed attempts. How can I prove that $z in N_K (x_0)$?
Part 2 — Closed Unit Ball
Since $Vert x - p Vert geq langle z,; x-x_0 rangle$ for all $x in mathbb{R}^n$. Choose $x = z + x_0$ to obtain:
$$
Vert z + x_0 - p Vert geq langle z,; z rangle = Vert z Vert^2
$$
Since $Vert z + x_0 - p Vert leq Vert z Vert + Vert x_0 - pVert$, then obtain:
$$
Vert z Vert leq 1 + frac{Vert x_0 - p Vert}{Vert z Vert}
$$
If $x_0 in text{bdry}(K)$, then $p = text{proj}_K (z + x_0) = x_0$ since $z$ is normal (I am unsure of this). Thus the previous inequality gives $Vert z Vert leq 1$.
I have no clue what to do for $x_0 in text{int}(K)$.
(Here $text{int}$ is interior, $text{bdry}$ is boundary).
I am very confused and have put in too much time without any results. Please, any insight is immensely appreciated.
general-topology proof-verification convex-analysis convex-optimization convex-geometry
add a comment |
up vote
0
down vote
favorite
GIVEN
Let $K subset mathbb{R}^n$ be a nonempty, closed and convex set. The associated distance function is $d_K$. Find the subdifferential $partial d_K(x_0)$ for all $x_0 in K$.
USEFUL DEFINITIONS
Convex Subdifferential
The subdifferential of a proper, lower semi-continuous, and convex function $f$ at a point $x_0 in text{dom}(f)$ is:
$$
partial f (x_0) = big{ z in mathbb{R}^n : f(x) geq f(x_0) + langle z, x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Normal Cone
The normal cone of a nonempty, closed, and convex set $K$ at a point $x_0 in K$ is:
$$
N_K(x_0)=big{ z:langle z, ;x-x_0 rangle leq 0,; forall x in Kbig}
$$
CONTEXT
After some research I have found out that this is a well-known result: $partial d_K (x_0) = bar{B}cap N_K (x_0)$ for all $x_0 in K$. Here $bar{B}$ refers to the closed unit ball.
Note that this is true for Banach spaces. I am unsure of the result in my case. I also cannot use infimal convolution as I have not learned it yet.
However, I could not find any proof of it anywhere, and thus I am trying to prove that $Vert z Vert leq 1$ (giving $z in bar{B}$) and that $langle z, ; x - x_0 rangle leq 0$ (giving $z in N_K(x_0)$). Here $z in partial d_K (x_0)$.
ATTEMPT
Let $x_0$ be a fixed point in $K$.
The distance function $d_K(x) = inf_{y in K} Vert x - yVert$ over a convex set $K$ is convex, proper and lower semi-continuous. Since $x_0 in K$ then $d_K (x_0) = 0$.
Using the above definition, obtain:
$$
partial d_K (x_0) = big{ zeta in mathbb{R}^n : d_K(x) geq langle z, ; x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Part 1 — Normal Cone
Case 1: $x in K$
Here, $d_K (x) = 0$ and the above equation gives $partial d_K(x_0)=N_K (x_0)$.
Case 2: $x notin K$
Here, $d_K (x) = Vert x - pVert > 0$, where $p$ is the projection of $x$ onto $K$.
I am completely stuck here and have tried multiple failed attempts. How can I prove that $z in N_K (x_0)$?
Part 2 — Closed Unit Ball
Since $Vert x - p Vert geq langle z,; x-x_0 rangle$ for all $x in mathbb{R}^n$. Choose $x = z + x_0$ to obtain:
$$
Vert z + x_0 - p Vert geq langle z,; z rangle = Vert z Vert^2
$$
Since $Vert z + x_0 - p Vert leq Vert z Vert + Vert x_0 - pVert$, then obtain:
$$
Vert z Vert leq 1 + frac{Vert x_0 - p Vert}{Vert z Vert}
$$
If $x_0 in text{bdry}(K)$, then $p = text{proj}_K (z + x_0) = x_0$ since $z$ is normal (I am unsure of this). Thus the previous inequality gives $Vert z Vert leq 1$.
I have no clue what to do for $x_0 in text{int}(K)$.
(Here $text{int}$ is interior, $text{bdry}$ is boundary).
I am very confused and have put in too much time without any results. Please, any insight is immensely appreciated.
general-topology proof-verification convex-analysis convex-optimization convex-geometry
1
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
1
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
1
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
1
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19
add a comment |
up vote
0
down vote
favorite
up vote
0
down vote
favorite
GIVEN
Let $K subset mathbb{R}^n$ be a nonempty, closed and convex set. The associated distance function is $d_K$. Find the subdifferential $partial d_K(x_0)$ for all $x_0 in K$.
USEFUL DEFINITIONS
Convex Subdifferential
The subdifferential of a proper, lower semi-continuous, and convex function $f$ at a point $x_0 in text{dom}(f)$ is:
$$
partial f (x_0) = big{ z in mathbb{R}^n : f(x) geq f(x_0) + langle z, x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Normal Cone
The normal cone of a nonempty, closed, and convex set $K$ at a point $x_0 in K$ is:
$$
N_K(x_0)=big{ z:langle z, ;x-x_0 rangle leq 0,; forall x in Kbig}
$$
CONTEXT
After some research I have found out that this is a well-known result: $partial d_K (x_0) = bar{B}cap N_K (x_0)$ for all $x_0 in K$. Here $bar{B}$ refers to the closed unit ball.
Note that this is true for Banach spaces. I am unsure of the result in my case. I also cannot use infimal convolution as I have not learned it yet.
However, I could not find any proof of it anywhere, and thus I am trying to prove that $Vert z Vert leq 1$ (giving $z in bar{B}$) and that $langle z, ; x - x_0 rangle leq 0$ (giving $z in N_K(x_0)$). Here $z in partial d_K (x_0)$.
ATTEMPT
Let $x_0$ be a fixed point in $K$.
The distance function $d_K(x) = inf_{y in K} Vert x - yVert$ over a convex set $K$ is convex, proper and lower semi-continuous. Since $x_0 in K$ then $d_K (x_0) = 0$.
Using the above definition, obtain:
$$
partial d_K (x_0) = big{ zeta in mathbb{R}^n : d_K(x) geq langle z, ; x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Part 1 — Normal Cone
Case 1: $x in K$
Here, $d_K (x) = 0$ and the above equation gives $partial d_K(x_0)=N_K (x_0)$.
Case 2: $x notin K$
Here, $d_K (x) = Vert x - pVert > 0$, where $p$ is the projection of $x$ onto $K$.
I am completely stuck here and have tried multiple failed attempts. How can I prove that $z in N_K (x_0)$?
Part 2 — Closed Unit Ball
Since $Vert x - p Vert geq langle z,; x-x_0 rangle$ for all $x in mathbb{R}^n$. Choose $x = z + x_0$ to obtain:
$$
Vert z + x_0 - p Vert geq langle z,; z rangle = Vert z Vert^2
$$
Since $Vert z + x_0 - p Vert leq Vert z Vert + Vert x_0 - pVert$, then obtain:
$$
Vert z Vert leq 1 + frac{Vert x_0 - p Vert}{Vert z Vert}
$$
If $x_0 in text{bdry}(K)$, then $p = text{proj}_K (z + x_0) = x_0$ since $z$ is normal (I am unsure of this). Thus the previous inequality gives $Vert z Vert leq 1$.
I have no clue what to do for $x_0 in text{int}(K)$.
(Here $text{int}$ is interior, $text{bdry}$ is boundary).
I am very confused and have put in too much time without any results. Please, any insight is immensely appreciated.
general-topology proof-verification convex-analysis convex-optimization convex-geometry
GIVEN
Let $K subset mathbb{R}^n$ be a nonempty, closed and convex set. The associated distance function is $d_K$. Find the subdifferential $partial d_K(x_0)$ for all $x_0 in K$.
USEFUL DEFINITIONS
Convex Subdifferential
The subdifferential of a proper, lower semi-continuous, and convex function $f$ at a point $x_0 in text{dom}(f)$ is:
$$
partial f (x_0) = big{ z in mathbb{R}^n : f(x) geq f(x_0) + langle z, x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Normal Cone
The normal cone of a nonempty, closed, and convex set $K$ at a point $x_0 in K$ is:
$$
N_K(x_0)=big{ z:langle z, ;x-x_0 rangle leq 0,; forall x in Kbig}
$$
CONTEXT
After some research I have found out that this is a well-known result: $partial d_K (x_0) = bar{B}cap N_K (x_0)$ for all $x_0 in K$. Here $bar{B}$ refers to the closed unit ball.
Note that this is true for Banach spaces. I am unsure of the result in my case. I also cannot use infimal convolution as I have not learned it yet.
However, I could not find any proof of it anywhere, and thus I am trying to prove that $Vert z Vert leq 1$ (giving $z in bar{B}$) and that $langle z, ; x - x_0 rangle leq 0$ (giving $z in N_K(x_0)$). Here $z in partial d_K (x_0)$.
ATTEMPT
Let $x_0$ be a fixed point in $K$.
The distance function $d_K(x) = inf_{y in K} Vert x - yVert$ over a convex set $K$ is convex, proper and lower semi-continuous. Since $x_0 in K$ then $d_K (x_0) = 0$.
Using the above definition, obtain:
$$
partial d_K (x_0) = big{ zeta in mathbb{R}^n : d_K(x) geq langle z, ; x - x_0 rangle , ; forall x in mathbb{R}^n big}
$$
Part 1 — Normal Cone
Case 1: $x in K$
Here, $d_K (x) = 0$ and the above equation gives $partial d_K(x_0)=N_K (x_0)$.
Case 2: $x notin K$
Here, $d_K (x) = Vert x - pVert > 0$, where $p$ is the projection of $x$ onto $K$.
I am completely stuck here and have tried multiple failed attempts. How can I prove that $z in N_K (x_0)$?
Part 2 — Closed Unit Ball
Since $Vert x - p Vert geq langle z,; x-x_0 rangle$ for all $x in mathbb{R}^n$. Choose $x = z + x_0$ to obtain:
$$
Vert z + x_0 - p Vert geq langle z,; z rangle = Vert z Vert^2
$$
Since $Vert z + x_0 - p Vert leq Vert z Vert + Vert x_0 - pVert$, then obtain:
$$
Vert z Vert leq 1 + frac{Vert x_0 - p Vert}{Vert z Vert}
$$
If $x_0 in text{bdry}(K)$, then $p = text{proj}_K (z + x_0) = x_0$ since $z$ is normal (I am unsure of this). Thus the previous inequality gives $Vert z Vert leq 1$.
I have no clue what to do for $x_0 in text{int}(K)$.
(Here $text{int}$ is interior, $text{bdry}$ is boundary).
I am very confused and have put in too much time without any results. Please, any insight is immensely appreciated.
general-topology proof-verification convex-analysis convex-optimization convex-geometry
general-topology proof-verification convex-analysis convex-optimization convex-geometry
asked Nov 22 at 15:18
ex.nihil
17610
17610
1
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
1
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
1
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
1
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19
add a comment |
1
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
1
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
1
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
1
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19
1
1
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
1
1
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
1
1
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
1
1
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19
add a comment |
2 Answers
2
active
oldest
votes
up vote
1
down vote
accepted
If $x_0$ belongs to the boundary of $K$, then the subdifferential requested is the intersection of $N_K(x_0)$ and the closed unit ball.
If $x_0$ belongs to the interior of $K$, then it is ${0}$.
For completeness, if $x_0$ is outside $K$, then it is $(x_0-P_Kx_0)/d_K(x_0)$.
For proofs, see Bauschke-Combettes Convex Analysis and Monotone Operator Theory in Hilbert Spaces, second edition, Example 16.62.
add a comment |
up vote
0
down vote
Hint: If $x notin K$ then $d_K$ is differentiable at $x$.
Let $bar{x} in K$ is the unique point projection of $x$ on $K$
Try to prove that it is differentiable at $x$ using definition of differentiability with $ D d_k (x) = frac{x- bar{x}}{ | x- bar{x} |} $.
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
|
show 5 more comments
2 Answers
2
active
oldest
votes
2 Answers
2
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
1
down vote
accepted
If $x_0$ belongs to the boundary of $K$, then the subdifferential requested is the intersection of $N_K(x_0)$ and the closed unit ball.
If $x_0$ belongs to the interior of $K$, then it is ${0}$.
For completeness, if $x_0$ is outside $K$, then it is $(x_0-P_Kx_0)/d_K(x_0)$.
For proofs, see Bauschke-Combettes Convex Analysis and Monotone Operator Theory in Hilbert Spaces, second edition, Example 16.62.
add a comment |
up vote
1
down vote
accepted
If $x_0$ belongs to the boundary of $K$, then the subdifferential requested is the intersection of $N_K(x_0)$ and the closed unit ball.
If $x_0$ belongs to the interior of $K$, then it is ${0}$.
For completeness, if $x_0$ is outside $K$, then it is $(x_0-P_Kx_0)/d_K(x_0)$.
For proofs, see Bauschke-Combettes Convex Analysis and Monotone Operator Theory in Hilbert Spaces, second edition, Example 16.62.
add a comment |
up vote
1
down vote
accepted
up vote
1
down vote
accepted
If $x_0$ belongs to the boundary of $K$, then the subdifferential requested is the intersection of $N_K(x_0)$ and the closed unit ball.
If $x_0$ belongs to the interior of $K$, then it is ${0}$.
For completeness, if $x_0$ is outside $K$, then it is $(x_0-P_Kx_0)/d_K(x_0)$.
For proofs, see Bauschke-Combettes Convex Analysis and Monotone Operator Theory in Hilbert Spaces, second edition, Example 16.62.
If $x_0$ belongs to the boundary of $K$, then the subdifferential requested is the intersection of $N_K(x_0)$ and the closed unit ball.
If $x_0$ belongs to the interior of $K$, then it is ${0}$.
For completeness, if $x_0$ is outside $K$, then it is $(x_0-P_Kx_0)/d_K(x_0)$.
For proofs, see Bauschke-Combettes Convex Analysis and Monotone Operator Theory in Hilbert Spaces, second edition, Example 16.62.
answered Nov 23 at 6:24
max_zorn
3,28361328
3,28361328
add a comment |
add a comment |
up vote
0
down vote
Hint: If $x notin K$ then $d_K$ is differentiable at $x$.
Let $bar{x} in K$ is the unique point projection of $x$ on $K$
Try to prove that it is differentiable at $x$ using definition of differentiability with $ D d_k (x) = frac{x- bar{x}}{ | x- bar{x} |} $.
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
|
show 5 more comments
up vote
0
down vote
Hint: If $x notin K$ then $d_K$ is differentiable at $x$.
Let $bar{x} in K$ is the unique point projection of $x$ on $K$
Try to prove that it is differentiable at $x$ using definition of differentiability with $ D d_k (x) = frac{x- bar{x}}{ | x- bar{x} |} $.
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
|
show 5 more comments
up vote
0
down vote
up vote
0
down vote
Hint: If $x notin K$ then $d_K$ is differentiable at $x$.
Let $bar{x} in K$ is the unique point projection of $x$ on $K$
Try to prove that it is differentiable at $x$ using definition of differentiability with $ D d_k (x) = frac{x- bar{x}}{ | x- bar{x} |} $.
Hint: If $x notin K$ then $d_K$ is differentiable at $x$.
Let $bar{x} in K$ is the unique point projection of $x$ on $K$
Try to prove that it is differentiable at $x$ using definition of differentiability with $ D d_k (x) = frac{x- bar{x}}{ | x- bar{x} |} $.
answered Nov 22 at 17:43
Red shoes
4,656621
4,656621
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
|
show 5 more comments
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
1
1
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
Not relevant. Read the question. He is explicitly asking for $x_0in K$.
– Federico
Nov 22 at 17:52
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
He said he got stuck in case two, I gave him a big hint. what I'm saying is that in case 2 $partial d_k (x_0) = { frac{x_ 0 - bar{x}}{ | x_0 - bar{x} |} } $
– Red shoes
Nov 22 at 18:12
1
1
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
But if $xnotin K$ then $N_K(x)$ has nothing to do with $partial d_K(x)$
– Federico
Nov 22 at 18:16
1
1
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
I mean, of course what you say is correct, I just don't see what it has to do with "Find the convex subdifferential $partial d_K$ of the distance function $d_K$ associated to a convex set $K$ at in-set points $x_0 in K$."
– Federico
Nov 22 at 18:18
1
1
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
Actually, under the section ATTEMPT, he reaches a wrong conclusion also in Case 1.
– Federico
Nov 22 at 18:28
|
show 5 more comments
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmath.stackexchange.com%2fquestions%2f3009263%2ffind-the-convex-subdifferential-partial-d-k-of-the-distance-function-d-k-as%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
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
1
Can you prove at least an inclusion? $partial d_K(x_0)subset N_K(x_0)$ is immediate.
– Federico
Nov 22 at 17:54
1
$partial d_K(x_0)subset B_1$ is also clear because $d_K$ is $1$-Lipschitz.
– Federico
Nov 22 at 17:55
Of course you can just work with $x_0inpartial K$, because if $x_0inmathop{mathrm{int}}(K)$ then both $partial d_K(x_0)$ and $N_K(x_0)$ are empty.
– Federico
Nov 22 at 17:57
1
@Federico Empty !!!!! if $x_0 in int(K)$ then $ partial d_k (x_0) = N_K (x_0) = {0} $
– Red shoes
Nov 22 at 18:18
1
@Redshoes right, of course, my bad; I rushed over it
– Federico
Nov 22 at 18:19