How to integrate over the standard $n$-simplex directly in $mathbb{R}^{n+1}$?
up vote
3
down vote
favorite
Definition of standard $n$-simplex
The standard $n$-simplex contains all points $vec{x} in mathbb{R}^{n + 1}$ such that $0 le x_i le 1$ and $vec{x} cdot vec{1} = 1$.
My definition of the area of $n$-simplex
$$A_{n} = frac{sqrt{n+1}}{n!}.$$
The standard $2$-simplex is an equilateral triangle with side length $sqrt{2}$ and vertices at $(1, 0, 0)$, $(0, 1, 0)$, and $(0, 0, 1)$. The area is $A_{2} = dfrac{sqrt{3}}2$.
The standard $1$-simplex is a line with vertices at $(1, 0)$ and $(0, 1)$. The length is $A_{1} = sqrt{2}$
I found the area of the standard $n$-simplex by first moving the simplex from $mathbb{R}^{n+1}$ to $mathbb{R}^{n}$.
Area of standard simplex
Question
How to directly integrate over the standard $n$-simplex with the coordinates of $mathbb{R}^{n+1}$?
I can't use the change of coordinate formula because the Jacobian of the transformation from $mathbb{R}^{n}$ to $mathbb{R}^{n+1}$ is rectangular.
Thanks.
calculus linear-algebra integration geometry multivariable-calculus
add a comment |
up vote
3
down vote
favorite
Definition of standard $n$-simplex
The standard $n$-simplex contains all points $vec{x} in mathbb{R}^{n + 1}$ such that $0 le x_i le 1$ and $vec{x} cdot vec{1} = 1$.
My definition of the area of $n$-simplex
$$A_{n} = frac{sqrt{n+1}}{n!}.$$
The standard $2$-simplex is an equilateral triangle with side length $sqrt{2}$ and vertices at $(1, 0, 0)$, $(0, 1, 0)$, and $(0, 0, 1)$. The area is $A_{2} = dfrac{sqrt{3}}2$.
The standard $1$-simplex is a line with vertices at $(1, 0)$ and $(0, 1)$. The length is $A_{1} = sqrt{2}$
I found the area of the standard $n$-simplex by first moving the simplex from $mathbb{R}^{n+1}$ to $mathbb{R}^{n}$.
Area of standard simplex
Question
How to directly integrate over the standard $n$-simplex with the coordinates of $mathbb{R}^{n+1}$?
I can't use the change of coordinate formula because the Jacobian of the transformation from $mathbb{R}^{n}$ to $mathbb{R}^{n+1}$ is rectangular.
Thanks.
calculus linear-algebra integration geometry multivariable-calculus
add a comment |
up vote
3
down vote
favorite
up vote
3
down vote
favorite
Definition of standard $n$-simplex
The standard $n$-simplex contains all points $vec{x} in mathbb{R}^{n + 1}$ such that $0 le x_i le 1$ and $vec{x} cdot vec{1} = 1$.
My definition of the area of $n$-simplex
$$A_{n} = frac{sqrt{n+1}}{n!}.$$
The standard $2$-simplex is an equilateral triangle with side length $sqrt{2}$ and vertices at $(1, 0, 0)$, $(0, 1, 0)$, and $(0, 0, 1)$. The area is $A_{2} = dfrac{sqrt{3}}2$.
The standard $1$-simplex is a line with vertices at $(1, 0)$ and $(0, 1)$. The length is $A_{1} = sqrt{2}$
I found the area of the standard $n$-simplex by first moving the simplex from $mathbb{R}^{n+1}$ to $mathbb{R}^{n}$.
Area of standard simplex
Question
How to directly integrate over the standard $n$-simplex with the coordinates of $mathbb{R}^{n+1}$?
I can't use the change of coordinate formula because the Jacobian of the transformation from $mathbb{R}^{n}$ to $mathbb{R}^{n+1}$ is rectangular.
Thanks.
calculus linear-algebra integration geometry multivariable-calculus
Definition of standard $n$-simplex
The standard $n$-simplex contains all points $vec{x} in mathbb{R}^{n + 1}$ such that $0 le x_i le 1$ and $vec{x} cdot vec{1} = 1$.
My definition of the area of $n$-simplex
$$A_{n} = frac{sqrt{n+1}}{n!}.$$
The standard $2$-simplex is an equilateral triangle with side length $sqrt{2}$ and vertices at $(1, 0, 0)$, $(0, 1, 0)$, and $(0, 0, 1)$. The area is $A_{2} = dfrac{sqrt{3}}2$.
The standard $1$-simplex is a line with vertices at $(1, 0)$ and $(0, 1)$. The length is $A_{1} = sqrt{2}$
I found the area of the standard $n$-simplex by first moving the simplex from $mathbb{R}^{n+1}$ to $mathbb{R}^{n}$.
Area of standard simplex
Question
How to directly integrate over the standard $n$-simplex with the coordinates of $mathbb{R}^{n+1}$?
I can't use the change of coordinate formula because the Jacobian of the transformation from $mathbb{R}^{n}$ to $mathbb{R}^{n+1}$ is rectangular.
Thanks.
calculus linear-algebra integration geometry multivariable-calculus
calculus linear-algebra integration geometry multivariable-calculus
edited 2 days ago
Batominovski
30.8k23187
30.8k23187
asked Nov 13 at 4:36
R zu
32910
32910
add a comment |
add a comment |
1 Answer
1
active
oldest
votes
up vote
3
down vote
accepted
Let $Delta^nsubseteq mathbb{R}^{n+1}$ be the standard $n$-simplex: $$Delta^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}_{geq 0}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}.$$ An arbitrary function $f:Delta^nto mathbb{R}$ can be considered as a function in $n$ free variables. That is, there exists a unique function $F:Sigma^nto mathbb{R}$ such that
$$f(x_0,x_1,x_2,ldots,x_n)=F(x_1,x_2,ldots,x_n)$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$, where $$Sigma^n:=big{(x_1,x_2,ldots,x_n)subseteqmathbb{R}_{geq 0}^n,big|,x_1+x_2+ldots+x_nleq1big}subseteq mathbb{R}^n,.$$ In other words, $F$ is given by $$F(x_1,x_2,ldots,x_n):=fleft(1-x_1-x_2-ldots-x_n,x_1,x_2,ldots,x_nright),,$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$. That is, if $nu_n$ is the volume measure on $Delta^n$ and $lambda_n$ is the Lebesgue measure on $mathbb{R}^n$, then
$$int_{Delta^n},f,text{d}nu_n=sqrt{n+1},int_{Sigma^n},F,text{d}lambda_n,.$$
This is because $$text{d}nu_n(x_0,x_1,x_2,ldots,x_n)=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)$$
for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$ (this can be proven using the Jacobian determinant). In particular, if $fequiv 1$ so that $Fequiv 1$, then we get
$$text{vol}_nleft(Delta^nright)=sqrt{n+1},text{vol}_nleft(Sigma^nright)=frac{sqrt{n+1}}{n!},,$$
where $text{vol}_n$ is the $n$-dimensional volume.
To clarify some points, first note that there exists an isometry from the affine hyperplane $$H^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}$$
to $mathbb{R}^n$. We can take the isometry to be the unique affine map $varphi:H^ntomathbb{R}^{n}$ which sends $e_0,e_1,e_2,ldots,e_ninmathbb{R}^{n+1}$ to $$0,E_1+alpha_n, E,E_2+alpha_n, E,ldots,E_n+alpha_n, Einmathbb{R}^n,,$$ respectively, where $e_0,e_1,e_2,ldots,e_n$ are standard basis vectors of $mathbb{R}^{n+1}$, $E_1,E_2,ldots,E_n$ are standard basis vectors of $mathbb{R}^n$, $E:=E_1+E_2+ldots+E_n$, and
$$alpha_n:=frac{sqrt{n+1}-1}{n},.$$
Write $E_0:=-alpha_n,E$. Let $T:mathbb{R}^ntomathbb{R}^n$ be the unique linear transformation that sends $$E_1,E_2,ldots,E_ntext{ to }E_1-E_0,E_2-E_0,ldots,E_n-E_0,,$$ respectively. Prove that $$det(T)=1+n,alpha_n=sqrt{n+1},.$$
The volume measure $nu_n$ on $Delta^n$ is inherited from the volume measure on $H^n$, which is the pullback $varphi^*lambda_n$ of $lambda_n$. However, since $T$ maps the extreme points $0,E_1,E_2,ldots,E_n$ of $Sigma^n$ to the extreme points $0,E_1-E_0,E_2-E_0,ldots,E_n-E_0$ of $varphi(H^n)$, it is simpler to take the integral on $Sigma^n$, using the Change-of-Variables Theorem. That is,
$$begin{align}text{d}nu_n(x_0,x_1,ldots,x_n)&=text{d}(varphi^*lambda_n)left(x_0,x_1,ldots,x_nright)\&=text{d}lambda_nbig(varphileft(x_0,x_1,ldots,x_nright)big)\
&=text{d}lambda_nleft(x_1+alpha_n,sum_{i=1}^n,x_i,x_2+alpha_n,sum_{i=1}^n,x_i,ldots,x_n+alpha_n,sum_{i=1}^n,x_iright)
\&=text{d}lambda_nbig(T(x_1,x_2,ldots,x_n)big)\&=text{d}(T^*lambda_n)(x_1,x_2,ldots,x_n)\&=det(T),text{d}lambda_n(x_1,x_2,ldots,x_n)\&=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)end{align}$$
for all $(x_0,x_1,ldots,x_n)inDelta^n$.
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
add a comment |
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
3
down vote
accepted
Let $Delta^nsubseteq mathbb{R}^{n+1}$ be the standard $n$-simplex: $$Delta^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}_{geq 0}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}.$$ An arbitrary function $f:Delta^nto mathbb{R}$ can be considered as a function in $n$ free variables. That is, there exists a unique function $F:Sigma^nto mathbb{R}$ such that
$$f(x_0,x_1,x_2,ldots,x_n)=F(x_1,x_2,ldots,x_n)$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$, where $$Sigma^n:=big{(x_1,x_2,ldots,x_n)subseteqmathbb{R}_{geq 0}^n,big|,x_1+x_2+ldots+x_nleq1big}subseteq mathbb{R}^n,.$$ In other words, $F$ is given by $$F(x_1,x_2,ldots,x_n):=fleft(1-x_1-x_2-ldots-x_n,x_1,x_2,ldots,x_nright),,$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$. That is, if $nu_n$ is the volume measure on $Delta^n$ and $lambda_n$ is the Lebesgue measure on $mathbb{R}^n$, then
$$int_{Delta^n},f,text{d}nu_n=sqrt{n+1},int_{Sigma^n},F,text{d}lambda_n,.$$
This is because $$text{d}nu_n(x_0,x_1,x_2,ldots,x_n)=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)$$
for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$ (this can be proven using the Jacobian determinant). In particular, if $fequiv 1$ so that $Fequiv 1$, then we get
$$text{vol}_nleft(Delta^nright)=sqrt{n+1},text{vol}_nleft(Sigma^nright)=frac{sqrt{n+1}}{n!},,$$
where $text{vol}_n$ is the $n$-dimensional volume.
To clarify some points, first note that there exists an isometry from the affine hyperplane $$H^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}$$
to $mathbb{R}^n$. We can take the isometry to be the unique affine map $varphi:H^ntomathbb{R}^{n}$ which sends $e_0,e_1,e_2,ldots,e_ninmathbb{R}^{n+1}$ to $$0,E_1+alpha_n, E,E_2+alpha_n, E,ldots,E_n+alpha_n, Einmathbb{R}^n,,$$ respectively, where $e_0,e_1,e_2,ldots,e_n$ are standard basis vectors of $mathbb{R}^{n+1}$, $E_1,E_2,ldots,E_n$ are standard basis vectors of $mathbb{R}^n$, $E:=E_1+E_2+ldots+E_n$, and
$$alpha_n:=frac{sqrt{n+1}-1}{n},.$$
Write $E_0:=-alpha_n,E$. Let $T:mathbb{R}^ntomathbb{R}^n$ be the unique linear transformation that sends $$E_1,E_2,ldots,E_ntext{ to }E_1-E_0,E_2-E_0,ldots,E_n-E_0,,$$ respectively. Prove that $$det(T)=1+n,alpha_n=sqrt{n+1},.$$
The volume measure $nu_n$ on $Delta^n$ is inherited from the volume measure on $H^n$, which is the pullback $varphi^*lambda_n$ of $lambda_n$. However, since $T$ maps the extreme points $0,E_1,E_2,ldots,E_n$ of $Sigma^n$ to the extreme points $0,E_1-E_0,E_2-E_0,ldots,E_n-E_0$ of $varphi(H^n)$, it is simpler to take the integral on $Sigma^n$, using the Change-of-Variables Theorem. That is,
$$begin{align}text{d}nu_n(x_0,x_1,ldots,x_n)&=text{d}(varphi^*lambda_n)left(x_0,x_1,ldots,x_nright)\&=text{d}lambda_nbig(varphileft(x_0,x_1,ldots,x_nright)big)\
&=text{d}lambda_nleft(x_1+alpha_n,sum_{i=1}^n,x_i,x_2+alpha_n,sum_{i=1}^n,x_i,ldots,x_n+alpha_n,sum_{i=1}^n,x_iright)
\&=text{d}lambda_nbig(T(x_1,x_2,ldots,x_n)big)\&=text{d}(T^*lambda_n)(x_1,x_2,ldots,x_n)\&=det(T),text{d}lambda_n(x_1,x_2,ldots,x_n)\&=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)end{align}$$
for all $(x_0,x_1,ldots,x_n)inDelta^n$.
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
add a comment |
up vote
3
down vote
accepted
Let $Delta^nsubseteq mathbb{R}^{n+1}$ be the standard $n$-simplex: $$Delta^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}_{geq 0}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}.$$ An arbitrary function $f:Delta^nto mathbb{R}$ can be considered as a function in $n$ free variables. That is, there exists a unique function $F:Sigma^nto mathbb{R}$ such that
$$f(x_0,x_1,x_2,ldots,x_n)=F(x_1,x_2,ldots,x_n)$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$, where $$Sigma^n:=big{(x_1,x_2,ldots,x_n)subseteqmathbb{R}_{geq 0}^n,big|,x_1+x_2+ldots+x_nleq1big}subseteq mathbb{R}^n,.$$ In other words, $F$ is given by $$F(x_1,x_2,ldots,x_n):=fleft(1-x_1-x_2-ldots-x_n,x_1,x_2,ldots,x_nright),,$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$. That is, if $nu_n$ is the volume measure on $Delta^n$ and $lambda_n$ is the Lebesgue measure on $mathbb{R}^n$, then
$$int_{Delta^n},f,text{d}nu_n=sqrt{n+1},int_{Sigma^n},F,text{d}lambda_n,.$$
This is because $$text{d}nu_n(x_0,x_1,x_2,ldots,x_n)=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)$$
for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$ (this can be proven using the Jacobian determinant). In particular, if $fequiv 1$ so that $Fequiv 1$, then we get
$$text{vol}_nleft(Delta^nright)=sqrt{n+1},text{vol}_nleft(Sigma^nright)=frac{sqrt{n+1}}{n!},,$$
where $text{vol}_n$ is the $n$-dimensional volume.
To clarify some points, first note that there exists an isometry from the affine hyperplane $$H^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}$$
to $mathbb{R}^n$. We can take the isometry to be the unique affine map $varphi:H^ntomathbb{R}^{n}$ which sends $e_0,e_1,e_2,ldots,e_ninmathbb{R}^{n+1}$ to $$0,E_1+alpha_n, E,E_2+alpha_n, E,ldots,E_n+alpha_n, Einmathbb{R}^n,,$$ respectively, where $e_0,e_1,e_2,ldots,e_n$ are standard basis vectors of $mathbb{R}^{n+1}$, $E_1,E_2,ldots,E_n$ are standard basis vectors of $mathbb{R}^n$, $E:=E_1+E_2+ldots+E_n$, and
$$alpha_n:=frac{sqrt{n+1}-1}{n},.$$
Write $E_0:=-alpha_n,E$. Let $T:mathbb{R}^ntomathbb{R}^n$ be the unique linear transformation that sends $$E_1,E_2,ldots,E_ntext{ to }E_1-E_0,E_2-E_0,ldots,E_n-E_0,,$$ respectively. Prove that $$det(T)=1+n,alpha_n=sqrt{n+1},.$$
The volume measure $nu_n$ on $Delta^n$ is inherited from the volume measure on $H^n$, which is the pullback $varphi^*lambda_n$ of $lambda_n$. However, since $T$ maps the extreme points $0,E_1,E_2,ldots,E_n$ of $Sigma^n$ to the extreme points $0,E_1-E_0,E_2-E_0,ldots,E_n-E_0$ of $varphi(H^n)$, it is simpler to take the integral on $Sigma^n$, using the Change-of-Variables Theorem. That is,
$$begin{align}text{d}nu_n(x_0,x_1,ldots,x_n)&=text{d}(varphi^*lambda_n)left(x_0,x_1,ldots,x_nright)\&=text{d}lambda_nbig(varphileft(x_0,x_1,ldots,x_nright)big)\
&=text{d}lambda_nleft(x_1+alpha_n,sum_{i=1}^n,x_i,x_2+alpha_n,sum_{i=1}^n,x_i,ldots,x_n+alpha_n,sum_{i=1}^n,x_iright)
\&=text{d}lambda_nbig(T(x_1,x_2,ldots,x_n)big)\&=text{d}(T^*lambda_n)(x_1,x_2,ldots,x_n)\&=det(T),text{d}lambda_n(x_1,x_2,ldots,x_n)\&=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)end{align}$$
for all $(x_0,x_1,ldots,x_n)inDelta^n$.
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
add a comment |
up vote
3
down vote
accepted
up vote
3
down vote
accepted
Let $Delta^nsubseteq mathbb{R}^{n+1}$ be the standard $n$-simplex: $$Delta^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}_{geq 0}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}.$$ An arbitrary function $f:Delta^nto mathbb{R}$ can be considered as a function in $n$ free variables. That is, there exists a unique function $F:Sigma^nto mathbb{R}$ such that
$$f(x_0,x_1,x_2,ldots,x_n)=F(x_1,x_2,ldots,x_n)$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$, where $$Sigma^n:=big{(x_1,x_2,ldots,x_n)subseteqmathbb{R}_{geq 0}^n,big|,x_1+x_2+ldots+x_nleq1big}subseteq mathbb{R}^n,.$$ In other words, $F$ is given by $$F(x_1,x_2,ldots,x_n):=fleft(1-x_1-x_2-ldots-x_n,x_1,x_2,ldots,x_nright),,$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$. That is, if $nu_n$ is the volume measure on $Delta^n$ and $lambda_n$ is the Lebesgue measure on $mathbb{R}^n$, then
$$int_{Delta^n},f,text{d}nu_n=sqrt{n+1},int_{Sigma^n},F,text{d}lambda_n,.$$
This is because $$text{d}nu_n(x_0,x_1,x_2,ldots,x_n)=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)$$
for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$ (this can be proven using the Jacobian determinant). In particular, if $fequiv 1$ so that $Fequiv 1$, then we get
$$text{vol}_nleft(Delta^nright)=sqrt{n+1},text{vol}_nleft(Sigma^nright)=frac{sqrt{n+1}}{n!},,$$
where $text{vol}_n$ is the $n$-dimensional volume.
To clarify some points, first note that there exists an isometry from the affine hyperplane $$H^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}$$
to $mathbb{R}^n$. We can take the isometry to be the unique affine map $varphi:H^ntomathbb{R}^{n}$ which sends $e_0,e_1,e_2,ldots,e_ninmathbb{R}^{n+1}$ to $$0,E_1+alpha_n, E,E_2+alpha_n, E,ldots,E_n+alpha_n, Einmathbb{R}^n,,$$ respectively, where $e_0,e_1,e_2,ldots,e_n$ are standard basis vectors of $mathbb{R}^{n+1}$, $E_1,E_2,ldots,E_n$ are standard basis vectors of $mathbb{R}^n$, $E:=E_1+E_2+ldots+E_n$, and
$$alpha_n:=frac{sqrt{n+1}-1}{n},.$$
Write $E_0:=-alpha_n,E$. Let $T:mathbb{R}^ntomathbb{R}^n$ be the unique linear transformation that sends $$E_1,E_2,ldots,E_ntext{ to }E_1-E_0,E_2-E_0,ldots,E_n-E_0,,$$ respectively. Prove that $$det(T)=1+n,alpha_n=sqrt{n+1},.$$
The volume measure $nu_n$ on $Delta^n$ is inherited from the volume measure on $H^n$, which is the pullback $varphi^*lambda_n$ of $lambda_n$. However, since $T$ maps the extreme points $0,E_1,E_2,ldots,E_n$ of $Sigma^n$ to the extreme points $0,E_1-E_0,E_2-E_0,ldots,E_n-E_0$ of $varphi(H^n)$, it is simpler to take the integral on $Sigma^n$, using the Change-of-Variables Theorem. That is,
$$begin{align}text{d}nu_n(x_0,x_1,ldots,x_n)&=text{d}(varphi^*lambda_n)left(x_0,x_1,ldots,x_nright)\&=text{d}lambda_nbig(varphileft(x_0,x_1,ldots,x_nright)big)\
&=text{d}lambda_nleft(x_1+alpha_n,sum_{i=1}^n,x_i,x_2+alpha_n,sum_{i=1}^n,x_i,ldots,x_n+alpha_n,sum_{i=1}^n,x_iright)
\&=text{d}lambda_nbig(T(x_1,x_2,ldots,x_n)big)\&=text{d}(T^*lambda_n)(x_1,x_2,ldots,x_n)\&=det(T),text{d}lambda_n(x_1,x_2,ldots,x_n)\&=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)end{align}$$
for all $(x_0,x_1,ldots,x_n)inDelta^n$.
Let $Delta^nsubseteq mathbb{R}^{n+1}$ be the standard $n$-simplex: $$Delta^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}_{geq 0}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}.$$ An arbitrary function $f:Delta^nto mathbb{R}$ can be considered as a function in $n$ free variables. That is, there exists a unique function $F:Sigma^nto mathbb{R}$ such that
$$f(x_0,x_1,x_2,ldots,x_n)=F(x_1,x_2,ldots,x_n)$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$, where $$Sigma^n:=big{(x_1,x_2,ldots,x_n)subseteqmathbb{R}_{geq 0}^n,big|,x_1+x_2+ldots+x_nleq1big}subseteq mathbb{R}^n,.$$ In other words, $F$ is given by $$F(x_1,x_2,ldots,x_n):=fleft(1-x_1-x_2-ldots-x_n,x_1,x_2,ldots,x_nright),,$$ for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$. That is, if $nu_n$ is the volume measure on $Delta^n$ and $lambda_n$ is the Lebesgue measure on $mathbb{R}^n$, then
$$int_{Delta^n},f,text{d}nu_n=sqrt{n+1},int_{Sigma^n},F,text{d}lambda_n,.$$
This is because $$text{d}nu_n(x_0,x_1,x_2,ldots,x_n)=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)$$
for all $(x_0,x_1,x_2,ldots,x_n)inDelta^n$ (this can be proven using the Jacobian determinant). In particular, if $fequiv 1$ so that $Fequiv 1$, then we get
$$text{vol}_nleft(Delta^nright)=sqrt{n+1},text{vol}_nleft(Sigma^nright)=frac{sqrt{n+1}}{n!},,$$
where $text{vol}_n$ is the $n$-dimensional volume.
To clarify some points, first note that there exists an isometry from the affine hyperplane $$H^n:=big{(x_0,x_1,x_2,ldots,x_n)inmathbb{R}^{n+1},big|,x_0+x_1+x_2+ldots+x_n=1big}$$
to $mathbb{R}^n$. We can take the isometry to be the unique affine map $varphi:H^ntomathbb{R}^{n}$ which sends $e_0,e_1,e_2,ldots,e_ninmathbb{R}^{n+1}$ to $$0,E_1+alpha_n, E,E_2+alpha_n, E,ldots,E_n+alpha_n, Einmathbb{R}^n,,$$ respectively, where $e_0,e_1,e_2,ldots,e_n$ are standard basis vectors of $mathbb{R}^{n+1}$, $E_1,E_2,ldots,E_n$ are standard basis vectors of $mathbb{R}^n$, $E:=E_1+E_2+ldots+E_n$, and
$$alpha_n:=frac{sqrt{n+1}-1}{n},.$$
Write $E_0:=-alpha_n,E$. Let $T:mathbb{R}^ntomathbb{R}^n$ be the unique linear transformation that sends $$E_1,E_2,ldots,E_ntext{ to }E_1-E_0,E_2-E_0,ldots,E_n-E_0,,$$ respectively. Prove that $$det(T)=1+n,alpha_n=sqrt{n+1},.$$
The volume measure $nu_n$ on $Delta^n$ is inherited from the volume measure on $H^n$, which is the pullback $varphi^*lambda_n$ of $lambda_n$. However, since $T$ maps the extreme points $0,E_1,E_2,ldots,E_n$ of $Sigma^n$ to the extreme points $0,E_1-E_0,E_2-E_0,ldots,E_n-E_0$ of $varphi(H^n)$, it is simpler to take the integral on $Sigma^n$, using the Change-of-Variables Theorem. That is,
$$begin{align}text{d}nu_n(x_0,x_1,ldots,x_n)&=text{d}(varphi^*lambda_n)left(x_0,x_1,ldots,x_nright)\&=text{d}lambda_nbig(varphileft(x_0,x_1,ldots,x_nright)big)\
&=text{d}lambda_nleft(x_1+alpha_n,sum_{i=1}^n,x_i,x_2+alpha_n,sum_{i=1}^n,x_i,ldots,x_n+alpha_n,sum_{i=1}^n,x_iright)
\&=text{d}lambda_nbig(T(x_1,x_2,ldots,x_n)big)\&=text{d}(T^*lambda_n)(x_1,x_2,ldots,x_n)\&=det(T),text{d}lambda_n(x_1,x_2,ldots,x_n)\&=sqrt{n+1},text{d}lambda_n(x_1,x_2,ldots,x_n)end{align}$$
for all $(x_0,x_1,ldots,x_n)inDelta^n$.
edited 2 days ago
answered Nov 13 at 9:32
Batominovski
30.8k23187
30.8k23187
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
add a comment |
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
1
1
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
What do you mean by "volume measure on $Delta^n$", and how does this strange square root $sqrt{n+1}$ come about? Name dropping does not suffice.
– Christian Blatter
Nov 13 at 17:03
1
1
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
@ChristianBlatter Here, I expanded my answer per your comment. I intentionally left the calculations to the OP. But I admit that the calculations are not straightforward at all. And I think your answer is good, so it should be undeleted, if you don't mind.
– Batominovski
Nov 13 at 19:56
add a comment |
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%2f2996301%2fhow-to-integrate-over-the-standard-n-simplex-directly-in-mathbbrn1%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