Questions tagged [gaussian]
For questions about the Gaussian probability distribution, its definition, properties and use.
549
questions
0
votes
1
answer
32
views
Two definitions of a Stochastic Process?
I have two supposedly equivalent definitions of a stochastic process.
A stochastic process is an indexed set of random variables. Specifically
$$
y = \{y(x) \; | \; x \in \mathcal{X}\}.
$$
...
- 651
0
votes
1
answer
27
views
Property $E[\mathbf{xx}^T] = \mathbf{\Sigma}-\mathbf{\mu\mu}^T$.
I'm reading through Principles of Machine Learning by Murphy for the purpose of understanding Gaussian Processes. For now I just want to understand the multivariate Gaussian distribution but I'm ...
- 651
1
vote
0
answers
16
views
Bound KL divergence between two distributions by KL divergence of two Gaussian mixture models
I'm trying to bound the KL divergence between two continuous random variables with the KL divergence between two Gaussian mixture approximations motivated by the fact that the Gaussian mixture model ...
- 824
0
votes
0
answers
16
views
Modeling a sequential Random Vector as a Random Processes
I'm studying random vectors. For a problem that I'm studying, I have a random vector that has a sequential nature. I know that modeling random variables as sequential processes is considered in the ...
2
votes
1
answer
36
views
projection of a non-zero mean Gaussian vector into a Ball
Let $d$ denote the dimension, $\mathbf{B}_d$ denote the ball of radius one in $\mathbb{R}^d$. For $x\in \mathbb{R}^d$ let $\Pi_{\mathbf{B}_d}(x) = \frac{x}{\max\{1,\|x\|_2\}}$.
Consider a fixed vector ...
- 561
0
votes
0
answers
32
views
How do I prove that the distribution of $x_t$ in $x_t=x_{t-1}e^{-μt}+ θ(1-e^{-μΔt}) + \int_{t-1}^t e^{-μ(t-s)}\sigma dB_s $ is a normal distribution?
$B_s$ is brownian motion. Because $\int_{t-1}^t e^{-μ(t-s)}\sigma dB_s $ has a Brownian component, it is normally distributed according to Taylor & Karin 1998 's Introduction to Stochastic ...
0
votes
0
answers
30
views
Derivation in paper Deep Neural Networks as Gaussian processes in ICLR 2018
I am trying to understand the derivation of the main equation in the seminal paper titled Deep Neural Networks as Gaussian processes (in ICLR 2018). Following is the equation number (7), which can be ...
0
votes
0
answers
22
views
What is the integral of a 2D Gaussian over a disk centered at the origin
Formal Statement
Let $G$ be a 2D symmetric Gaussian such that
$G(x, y; x_0, y_0, \sigma) = \exp\left(-\frac{\left( (x - x_0)^2 + (y - y_0)^2 \right)}{2 \sigma^2} \right)$,
for real $x, y, x_0, y_0$, ...
- 244
1
vote
0
answers
48
views
Solving Backward Heat Equation with a Backward Heat Kernel?
Let $D>0$ be a constant. Imagine we have the following forward heat conduction problem:
\begin{align*}
\begin{cases}
\partial_t u = D \partial_x^2u &, \quad (x,t) \in \mathbb{R} \times (0, \...
- 33
1
vote
1
answer
34
views
How do we compute an integral over a unit simplex?
Let $ n \ge 2 $ and $ T > n $ be integers.
The joint-distribution of eigenvalues in the Wishart ensemble subject to the underlying covariance matrix being equal to an identity matrix is given as ...
- 9,043
0
votes
0
answers
20
views
Jacobian in multidimensional Gaussian integral
For the first integral shown in:
reference for multidimensional gaussian integral
It is mentioned that the Jacobian is 1, how is this the case? By following the calculation I found it to be $det(diag(...
- 53
0
votes
0
answers
20
views
Hanson-Wright Inequality Lowering the Constant to Obtain a Factor of 2 Instead of 4
I'm reading the proof of Hanson-Wright Inequality from Rudelson & Vershynin's paper "Hanson-Wright inequality and sub-gaussian concentration". (https://arxiv.org/pdf/1306.2872.pdf)
I'm ...
- 73
0
votes
0
answers
19
views
Gaussian tail probability
Let $z(q)$ be the quaitle of a standard normal random variable $Z$, i.e., $z(q) = k$ when $Pr(Z\geq k) = q$. Then I would like to know why the following two results hold.
(a) If we hold $\alpha$ fixed,...
- 369
0
votes
1
answer
16
views
Rotation Invariance of the Distribution of $\ell_2$-norm square of Gaussian Vector
In a paper I'm reading I saw the argument below:
Let $G\sim\mathcal{N}(\mathbf{0},\mathbf{I})$ and $A$ is a square matrix with SVD $A=U\Sigma V^T$ where $\Sigma=diag(\sigma_1,\dots,\sigma_n)$. Then $|...
- 73
0
votes
2
answers
29
views
Finding the probability of an increment of a Brownian Bridge
I've been given a process $X_t=(1-t)B(\frac{t}{1-t}),0\leq t\leq 1$ where $B(\frac{t}{1-t})$ is a standard Brownian Motion.
So far, I have proven this is a Brownian bridge. Now I need to find $P(X_{4/...
- 9
0
votes
0
answers
23
views
Approximate 3D function that has certain points as peak
Given set of 3D points, how can I find a continuous function that has each point as the peak?
You can imagine making a terrain editor that the inputs are 3D points and the output is a smooth mountain ...
- 219
0
votes
0
answers
15
views
Frequency that a discrete brownian motion with drift process exceeds its previous maximum
This question relates to the frequency with which a discretely observed brownian motion exceeds its all time high.
Assume my retirement portfolio's value can be modelled as a process with normally ...
- 101
0
votes
0
answers
20
views
Computing a puzzling expectation with random variables and vectors.
Suppose the random variables $z_i $ are i.i.d. draw from the standard normal distribution $\mathcal{N}(0, 1)$, two vector $\mathbf{v}=(v_1, 0)^\top$, $v_1>0$, $\mathbf{w}=(w_1, w_2)^\top$, the ...
- 1
1
vote
0
answers
15
views
Derivation of the integral form of the numerator in the Bayesian inference equation???? (not on the denominator)
In the reference Gaussian processes: iterative sparse approximations by Csató, Lehel
(Csató, Lehel. Gaussian processes: iterative sparse approximations. Diss. Aston University, 2002), on page 20, ...
- 21
0
votes
0
answers
10
views
Number of estimated parameters in gaussian mixture model
Considering a gaussian mixture model with $n$ components, the model contains $n-1$ weight parameters to estimate and $2\cdot n$ parameters for mean and and variance to estimate. In total this model ...
- 21
0
votes
1
answer
43
views
Covariance function of the square of a Gaussian process
Let $X_t$, $t\in\mathbb{R}$, be a Gaussian process with mean $0$. Prove that $$
Cov(X_s^2,X_t^2)=2Cov(X_s,X_t)^2
$$
I don't know how to handle the squares here.
- 1,932
5
votes
0
answers
93
views
Farmer wants to know how wet their field is
Problem
A farmer wants a better understanding of rainfall on their field. Assuming rain falls randomly and with equal likelihood over the entire field, the farmer thinks they can model the volume of ...
- 191
0
votes
1
answer
28
views
Mean of Cubic and Quartic forms of Gaussians
I am trying to calculate the following means:
$$ E[ (x-\mu_k)b^T(x-\mu_l)(x-\mu_l)^T ] $$
$$ E[ (x-\mu_k)(x-\mu_k)^TA(x-\mu_l)(x-\mu_l)^T ] $$
Where x is some multivariate gaussian random variable.
I ...
- 25
0
votes
0
answers
14
views
Distribution of Complex Gaussian random variables after multiplication
I am very new to this topic, so please excuse me if the question is too naive.
I have the following equation:
$y= hx+n$ where $h \sim \mathbb{CN}(0,1)$ and $n \sim \mathbb{CN}(0,N_0I)$, $h,n$ and $y$ ...
- 47
1
vote
1
answer
34
views
Does a Gaussian random walk lead to a Gaussian distribution in the limit, even when the initial state is non-Gaussian?
Suppose we have an initial random variable, which is not Gaussian, but has mean $0$, std $1$. Now we add $N$ unit Gaussian variables to this initial random variable, and then renormalize to mean $0$, ...
- 65
0
votes
1
answer
52
views
If $X$, $Y$, and $Z$ are Gaussian, and $E[X|Y,Z]$ is linear in $Y,Z$, does it mean that $(X,Y,Z)$ is Gaussian?
Let $X$, $Y$, and $Z$ be Gaussian random variables that are not independent. This does not imply that $(X,Y,Z)$ is a Gaussian vector. But if we add the conditions that $E[X|Y,Z]$ is a linear function ...
- 2,040
1
vote
1
answer
28
views
How to calculate the properties of the resultant 2D Gaussian from the deconvolution of one anisotropic, 2D Gaussian with another?
Anisotropic, two-dimensional Gaussians, are described by the equation (from here):
$f(x,y)=A\exp \left( -(a(x-x_0)^2 + 2b(x-x_0)(y-y_0) + c(y-y_0)^2 )\right)$
where A is the amplitude of the peak, $...
- 11
1
vote
1
answer
42
views
Joint distribution of the Sum of gaussian random variables
Suppose $X_1,X_2,X_3$ are iid with distribution $\mathbb{N(\theta, \sigma^2)}$ and $Y_1 = X_1 + X_2$ and $Y_2 = X_2 + X_3$.
I need to find the joint distribution of $Y_1, Y_2$.
Here is my attempt:
...
- 47
2
votes
1
answer
53
views
Limit of expectation of maximum of standard normal distribution
I cannot find the way to show that
if $X_1,\dots,X_n$ are independent standard normal random variables, then
$$
\lim_{n \rightarrow \infty} \frac {\mathbb{E} \max_{i=1,\dots,n}X_i}{\sqrt{2\log n}} = 1
...
- 521
2
votes
1
answer
50
views
high dimensional probability vershynin Exercise 6.7.5
Lemma 6.7.4(Symmetrization with Gaussians). Let $X_1,\dots,X_N$ be independent, mean zero random vectors in a normed space. Let $g_1,\dots,g_N \sim N(0,1)$ be independent Gaussian random variables, ...
- 521
4
votes
1
answer
132
views
Comparing the Training Costs of Machine Learning Algorithm: A Mathematical Perspective
Recently, I was looking at the optimization functions required in training Kernel Based Methods compared to Neural Networks.
1) Kernel Methods:
For instance, I was looking at the optimization in ...
- 1,700
0
votes
1
answer
49
views
Why the scale factor of the product of two gaussian functions is the convolution of the same gaussian functions?
The product of two Gaussian functions
$$
f(x)=\frac{1}{\sqrt{2 \pi} \sigma_{f}} \exp\left(-\frac{x^{2}}{2 \sigma_{f}^{2}} \right) \quad \text { and } \quad g(x-y)=\frac{1}{\sqrt{2 \pi} \sigma_{g}} \...
- 255
2
votes
1
answer
28
views
Existence of non-volume preserving automorphisms of normal distribution
Let $X=\mathbb{R}^d$, let $p$ the standard normal distribution on $X$, with zero mean and identity covariance, and let $f:X \to X$ be a diffeomorphism that preserves this normal distribution. For ...
- 23
1
vote
1
answer
70
views
How to do regression for this stochastic process?
I wonder if there is some good way to determine the parameters for a random process. In my research work, I have a random process of the following form:
$$
X_t=\alpha_{1}G_{t-1}+\alpha_{2}G_{t-2}+...+\...
- 77
2
votes
1
answer
26
views
Is the distribution of a random variable normal if it is conditionally normal and the conditioning variable is normal?
Suppose we have a variable $\xi\sim N(\mu_2,\sigma_2)$ and another conditional variable $\eta | \xi \sim N(\mu_1,\sigma_1)$. Is $\eta$ normally distributed, and if so what is its mean and variance?
I ...
- 347
0
votes
0
answers
17
views
Representing rectangular function with Gaussian series
Is it possible to represent the rectangular function with an infinite series of scaled Gaussian functions?
https://en.wikipedia.org/wiki/Rectangular_function
Thank you,
Alex
- 53
2
votes
0
answers
30
views
Creeping for a Levy processes with infinite total mass
I am reading Theorem 7.11 from the book on Levy processes by Kyprianoy. I cannot intuitively grasp this:
In this theorem it is said that if the process has a Gaussian component then it creeps upwards. ...
- 21
0
votes
0
answers
16
views
PRML - Ex. 1.20. Behavior of probability mass & density of high-dimensional Gaussian distribution
I am following PRML by Bishop.
At exercise 1.20, there is a question I am not quite able to understand the significance of.
Would appreciate clarification around the same.
Question 1.20 -
In this ...
0
votes
1
answer
22
views
Distribution of multivariate gaussian under linear constraints
Let $Z_1 \ldots Z_n$ be i.i.d Gaussians and introduce a set of linear constraints 1 through k, where the $j$'s constraint is the a linear combination of the $Z$'s take a specific value, i.e.
$ \sum_{i=...
- 31
1
vote
1
answer
46
views
Boundedness of Gaussian Process for Borell-TIS inequality
I have a question regarding the Borell-TIS Theorem which presents a result for a Gaussian process. An assumption is that the Gaussian Process $\{f_t\}_{t\in T} $ is almost surely finite, i.e. $P(\sup_{...
- 13
0
votes
0
answers
17
views
Understanding well the definition of a Gaussian vector according to Le Gall's book.
I am reading the book "Brownian Motion, Martingales, and Stochastic Calculus", Jean-François Le Gall and I have a doubt regarding the Gaussian vector definition.
First, the enveiroment is $(\...
- 422
0
votes
1
answer
42
views
Computation of the Laplace transform of the Gaussian heat kernel
Currently I'm interested in the Laplace transform of the Gaussian heat kernel
$$
k_t(x):=\frac{1}{(4\pi t)^\frac{d}{2}}e ^{-\frac{|x|^2}{4t}}.
$$
Using the Laplace transform
$$
G_\lambda(x)=\int_0^\...
0
votes
0
answers
7
views
How to generate random number of non-standardized t distribution
Foreword: In the sparse Bayesian learning, the signal of interest $\mathbf{x}$ is assumed to follow conditional Gaussian distribution $p(\mathbf{x}|\boldsymbol{\alpha})=\prod_{i=1}^N\mathcal{N}(x_i|0,\...
- 339
3
votes
0
answers
57
views
Martingale with Gaussian marginals that is not jointly Gaussian
Let $X_0,X_1$ be random variables such that $E[X_1|X_0]=X_0$, $X_0 \sim \mathcal N (0, \sigma_0^2)$, and $X_1 \sim \mathcal N (0, \sigma_1^2)$ (notice that $\sigma_0^2 \leq \sigma_1^2$ by the ...
- 2,040
0
votes
0
answers
23
views
Uniform-Gaussian mixture
enter image description hereWould you help me finding the probability density function of a mixture of Uniform random variable with Gaussian random variable.
Z=x*y
Where x is Uniform distribution and ...
0
votes
0
answers
6
views
Quantiles of pairwise distance between samples from a multivariate Gaussian distribution?
Assume that I have a $D$-variate multivariate Gaussian distribution $\mathcal{N}\left(\mathbf{\mu},\mathbf{\Sigma}\right)$ with known mean $\mathbf{\mu}$ and covariance $\mathbf{\Sigma}$. I could draw ...
- 855
0
votes
0
answers
14
views
There's a metric to express the disagreement of predictions mathematically?
Let be $y = \{a_1, a_2, ..., a_n\}$ the distribution of predicted values given a variable $x$ such that $\sum_{k = 1}^K a_k = 1$. Let's consider now that for a given $x$ in fact we have $\{y\}_1^T$ ...
- 1
1
vote
1
answer
27
views
exponential power distribution
I see two different definitions of the exponential power distribution:
http://www.math.wm.edu/~leemis/chart/UDR/PDFs/Exponentialpower.pdf
$$f(x)= \exp(1-\exp(\lambda x^k)) \exp(\lambda x^k) \lambda k ...
- 669
2
votes
0
answers
93
views
A dimension-free upper bound for $\| X \|_{\infty}$ when $X \sim N(0,\Sigma)$
For a $p$-dimensional random vector $X \sim N(0,\Sigma)$, it is known that with high probability
$$
\| X \|_{\max} \le C \sqrt{\log p}
$$
where $C$ is some constant possibly depending on $\Sigma$. I ...
- 4,331
0
votes
0
answers
27
views
Question about PDF and probability in continuous distribution.
The integral of a pdf is the probability, and the probability of drawing a specific value from a continues distribution is zero. So why can we compute the probability of, say drawing $x=1$ from $N(0, ...
- 113