Numerical Optimization: Difference between revisions

From David's Wiki
Created page with "Numerical Optimization ==Line Search Methods== Basic idea: * For each iteration< ** Find a direction <math>p</math>. ** Then find a step length <math>\alpha</math> which d..."
 
 
(17 intermediate revisions by the same user not shown)
Line 2: Line 2:




==Convergence Rates==
[https://en.wikipedia.org/wiki/Rate_of_convergence Wikipedia]<br>
The rate of convergence is <math>\lim_{k \rightarrow \infty} \frac{|x_{k+1}-x^*|}{|x_{k}-x^*|^q}=L</math><br>
Iterative methods have the following convergence rates:
* If Q=1 and L=1 we have sublinear convergence.
* If Q=1 and <math>L\in(0,1)</math> we have linear convergence.
* If Q=1 and <math>L=0</math> we have superlinear convergence.
* If Q=2 and <math>L \leq \infty</math> we have quadratic convergence.


==Line Search Methods==
==Line Search Methods==
Basic idea:  
Basic idea:
* For each iteration<
* For each iteration
** Find a direction <math>p</math>.
** Find a direction <math>p</math>.
** Then find a step length <math>\alpha</math> which decreases <math>f</math>.
** Then find a step length <math>\alpha</math> which decreases <math>f</math>.
Line 16: Line 24:
** Find a region where you trust your model accurately represents your objective function.
** Find a region where you trust your model accurately represents your objective function.
** Take a step.
** Take a step.
<br>
Variables:
* <math>f</math> is your objective function.
* <math>m_k</math> is your quadratic model at iteration k.
* <math>x_k</math> is your point at iteration k.
Your model is <math>m_k(p) = f_k + g_k^T p + \frac{1}{2}p^T B_k p</math>
where <math>g_k = \nabla f(x_k)</math> and <math>B_k</math> is a symmetric matrix.<br>
At each iteration, you solve a constrained optimization subproblem to find the best step <math>p</math>.<br>
<math>\min_{p \in \mathbb{R}^n} m_k(p)</math> such that <math>\Vert p \Vert < \Delta_k </math>.
===Cauchy Point Algorithms===
The Cauchy point <math>p_k^c = \tau_k p_k^s</math><br>
where <math>p_k^s</math> minimizes the linear model in the trust region<br>
<math> p_k^s = \operatorname{argmin}_{p \in \mathbb{R}^n} f_k + g_k^Tp </math> s.t. <math>\Vert p \Vert \leq \Delta_k </math><br>
and <math>\tau_k</math> minimizes our quadratic model along the line <math>p_k^s</math>:<br>
<math>\tau_k = \operatorname{argmin}_{\tau \geq 0} m_k(\tau p_k^s)</math> s.t. <math>\Vert \tau p_k^s \Vert \leq \Delta_k </math><br>
This can be written explicitly as <math>p_k^c = - \tau_k \frac{\Delta_k}{\Vert g_K \Vert} g_k</math> where <math>\tau_k =
\begin{cases}
1 & \text{if }g_k^T B_k g_k \leq 0;\\
\min(\Vert g_k \Vert ^3/(\Delta_k g_k^T B_k  g_k), 1) & \text{otherwise}
\end{cases}
</math>
==Conjugate Gradient Methods==
[https://www.cs.cmu.edu/~15859n/RelatedWork/painless-conjugate-gradient.pdf Painless Conjugate Gradient]<br>
The goal is to solve <math>Ax=b</math> or equivalently <math>\min \phi(x)</math> where <math>\phi(x)=(1/2)x^T A x - b^Tx</math>.<br>
The practical CG algorithm will converge in at most <math>n</math> steps.
===Definitions===
Vectors <math>\{p_i\}</math> are conjugate w.r.t SPD matrix <math>A</math> if <math>p_i^T A p_j = 0</math> for <math>i \neq j</math>
;Notes
* <math>\{p_i\}</math> are linearly independent
===Algorithms===
Basic idea:<br>
* Find a conjugate direction <math>p_k</math>
* Take the step size <math>\alpha_k</math> which minimizes along <math>\phi(x)</math>
Practical CG method:<br>
{{ hidden | Code |
Below is code for the practical CG method.<br>
<pre>
x = x0;
r = A*x - b;
p = -r;
k = 1;
while(norm(r) > tole && k < size(x,2))
  Ap = A*p;
  alpha = (r'*r)/(p'*Ap);
  x = x + alpha * p;
  rn = r + alpha * Ap;
  beta = (rn'*rn)/(r'*r);
  p = -rn + beta * p;
  r = rn;
  k = k+1;
end
</pre>
}}
===Theorems===
Given a set of conjugate directions <math>\{p_0,...,p_{n-1}\}</math> we can generate a sequence of <math>x_k</math> with<br>
<math> x_{k+1}=x_k + \alpha_k p_k</math> where <math>\alpha_k = -\frac{r_k^T p_k}{p_k^T A p_k}</math> minimizes <math>\phi(x)</math><br>
; Theorem: For an <math>x_0</math>, the sequence <math>x_k</math> converges to the solution <math>x^*</math> in at most n steps.
; Theorem: At step k, the residual <math>r_k</math> is orthogonal to <math>\{p_0,...p_{k-1}\}</math> and the current iteration <math>x_{k}</math>
: minimizes <math>\phi(x)</math> over the set <math>\{x_0 + span(p_0,...,p_{k-1})\}</math>
===Convergence Rate===
The convergence rate can be estimated by:<br>
<math>\min_{P_k} \max_{1 \leq i \leq k} [1 + \lambda_i P_k(\lambda_i)]^2</math>


==Resources==
==Resources==
* [https://link.springer.com/book/10.1007%2F978-0-387-40065-5 Numerical Optimization by Nocedal and Wright (2006)<br>
* [https://link.springer.com/book/10.1007%2F978-0-387-40065-5 Numerical Optimization by Nocedal and Wright (2006)]

Latest revision as of 14:09, 14 November 2019

Numerical Optimization


Convergence Rates

Wikipedia
The rate of convergence is \(\displaystyle \lim_{k \rightarrow \infty} \frac{|x_{k+1}-x^*|}{|x_{k}-x^*|^q}=L\)
Iterative methods have the following convergence rates:

  • If Q=1 and L=1 we have sublinear convergence.
  • If Q=1 and \(\displaystyle L\in(0,1)\) we have linear convergence.
  • If Q=1 and \(\displaystyle L=0\) we have superlinear convergence.
  • If Q=2 and \(\displaystyle L \leq \infty\) we have quadratic convergence.

Line Search Methods

Basic idea:

  • For each iteration
    • Find a direction \(\displaystyle p\).
    • Then find a step length \(\displaystyle \alpha\) which decreases \(\displaystyle f\).
    • Take a step \(\displaystyle \alpha p\).

Trust Region Methods

Basic idea:

  • For each iteration
    • Assume a quadratic model of your objective function near a point.
    • Find a region where you trust your model accurately represents your objective function.
    • Take a step.


Variables:

  • \(\displaystyle f\) is your objective function.
  • \(\displaystyle m_k\) is your quadratic model at iteration k.
  • \(\displaystyle x_k\) is your point at iteration k.

Your model is \(\displaystyle m_k(p) = f_k + g_k^T p + \frac{1}{2}p^T B_k p\) where \(\displaystyle g_k = \nabla f(x_k)\) and \(\displaystyle B_k\) is a symmetric matrix.
At each iteration, you solve a constrained optimization subproblem to find the best step \(\displaystyle p\).
\(\displaystyle \min_{p \in \mathbb{R}^n} m_k(p)\) such that \(\displaystyle \Vert p \Vert \lt \Delta_k \).

Cauchy Point Algorithms

The Cauchy point \(\displaystyle p_k^c = \tau_k p_k^s\)
where \(\displaystyle p_k^s\) minimizes the linear model in the trust region
\(\displaystyle p_k^s = \operatorname{argmin}_{p \in \mathbb{R}^n} f_k + g_k^Tp \) s.t. \(\displaystyle \Vert p \Vert \leq \Delta_k \)
and \(\displaystyle \tau_k\) minimizes our quadratic model along the line \(\displaystyle p_k^s\):
\(\displaystyle \tau_k = \operatorname{argmin}_{\tau \geq 0} m_k(\tau p_k^s)\) s.t. \(\displaystyle \Vert \tau p_k^s \Vert \leq \Delta_k \)
This can be written explicitly as \(\displaystyle p_k^c = - \tau_k \frac{\Delta_k}{\Vert g_K \Vert} g_k\) where \(\displaystyle \tau_k = \begin{cases} 1 & \text{if }g_k^T B_k g_k \leq 0;\\ \min(\Vert g_k \Vert ^3/(\Delta_k g_k^T B_k g_k), 1) & \text{otherwise} \end{cases} \)

Conjugate Gradient Methods

Painless Conjugate Gradient
The goal is to solve \(\displaystyle Ax=b\) or equivalently \(\displaystyle \min \phi(x)\) where \(\displaystyle \phi(x)=(1/2)x^T A x - b^Tx\).
The practical CG algorithm will converge in at most \(\displaystyle n\) steps.

Definitions

Vectors \(\displaystyle \{p_i\}\) are conjugate w.r.t SPD matrix \(\displaystyle A\) if \(\displaystyle p_i^T A p_j = 0\) for \(\displaystyle i \neq j\)

Notes
  • \(\displaystyle \{p_i\}\) are linearly independent

Algorithms

Basic idea:

  • Find a conjugate direction \(\displaystyle p_k\)
  • Take the step size \(\displaystyle \alpha_k\) which minimizes along \(\displaystyle \phi(x)\)

Practical CG method:

Code

Below is code for the practical CG method.

x = x0;
r = A*x - b;
p = -r;
k = 1;
while(norm(r) > tole && k < size(x,2))
   Ap = A*p;
   alpha = (r'*r)/(p'*Ap);
   x = x + alpha * p;
   rn = r + alpha * Ap;
   beta = (rn'*rn)/(r'*r);
   p = -rn + beta * p;
   r = rn;
   k = k+1;
end

Theorems

Given a set of conjugate directions \(\displaystyle \{p_0,...,p_{n-1}\}\) we can generate a sequence of \(\displaystyle x_k\) with
\(\displaystyle x_{k+1}=x_k + \alpha_k p_k\) where \(\displaystyle \alpha_k = -\frac{r_k^T p_k}{p_k^T A p_k}\) minimizes \(\displaystyle \phi(x)\)

Theorem
For an \(\displaystyle x_0\), the sequence \(\displaystyle x_k\) converges to the solution \(\displaystyle x^*\) in at most n steps.
Theorem
At step k, the residual \(\displaystyle r_k\) is orthogonal to \(\displaystyle \{p_0,...p_{k-1}\}\) and the current iteration \(\displaystyle x_{k}\)
minimizes \(\displaystyle \phi(x)\) over the set \(\displaystyle \{x_0 + span(p_0,...,p_{k-1})\}\)

Convergence Rate

The convergence rate can be estimated by:
\(\displaystyle \min_{P_k} \max_{1 \leq i \leq k} [1 + \lambda_i P_k(\lambda_i)]^2\)

Resources