Questions tagged [quadratic-programming]
Questions on quadratic programming, the optimization of a quadratic objective function subject to affine constraints.
789 questions
0 votes
1 answer
43 views
How to model fixed costs that activate only when a quadratic variable is nonzero?
I have a constraint optimization problem where the Objectif function is the sum of quadratic functions of the form $$ C_i = a_i g_{i}^2 + b_i g_{i} + c_i $$ $\{g_i\}$ are the variables, $\{a_i\},\{b_i\...
7 votes
2 answers
395 views
Min and max of a quadratic in a rectangle
Let $R$ be some rectangle: $R=[a_1,b_1)\times \dots\times [a_n,b_n)\subset \mathbb{R}^n.$ Suppose $C$ is some $n\times n$ real positive-definite matrix. Is there a closed form for the min and max of $...
0 votes
0 answers
69 views
How does one differentiate the solution to a quadratic problem with respect to its parameters?
Context and setup I am developing a library for multi-objective optimization. The idea is that instead of doing gradient descent, we have a Jacobian matrix whose rows are the gradients of our ...
0 votes
1 answer
54 views
QP with additional inequality constraint. Is the solution the projection?
I appologise in advance if this is a simple question. I have been trying to check some references like Boyd's book, but i can't piece together the parts that I need to asnwer this question. I've tried ...
1 vote
0 answers
84 views
How to project point onto polyhedron?
I'm writing to ask for your advice on an optimisation problem (unfortunately I'm not an expert), namely: \begin{array}{ll} \min\limits_{x \in \mathbb{R}^n} & \frac{1}{2} \| \mathbf{x} - \mathbf{y}...
0 votes
0 answers
41 views
Will the new optimisation problem give the old solution?
In introducucing the support vector machines we have that we have $N$ vectors of dimension $p,$ $\{\textbf{x}_i\}_{i=1}^N$, we also have $\{y_i\}_{i=1}^N$ where $y_i\in\{-1,1\}$. If the two groups are ...
0 votes
0 answers
32 views
Cholesky decomposition of constrained quadratic program
Consider the following constrained quadratic optimization problem: $\min_{\mathbf{u}} \quad \mathbf{u}^T \mathbf{H} \mathbf{u} + \mathbf{u}^T \mathbf{g}$ subject to: $\mathbf{u}_{\text{min}} \leq \...
0 votes
0 answers
36 views
Reformulation of a minimization problem concerning projection
I'am having problems understanding the a step in the solution proposed by my teacher(s) for the problem to minimize $\frac{1}{2}(x-\bar{x})^\top Q(x-\bar{x}) + c^\top(x-\bar{x})$ subject to $Ax = b$, ...
5 votes
2 answers
200 views
Minimizing a pure quadratic function under box constraint
I currently have the following problem to solve : $$ \underset{\mathbf{x}\geq \mathbf{1}}{\min} \quad \mathbf{x}^T\mathbf{Q}\mathbf{x}. $$ Here $\mathbf{x} \geq \mathbf{1}$ refers to elementwise ...
0 votes
1 answer
106 views
Clarifying different forms of the Dual of a quadratic Problem
I am trying to dualize a QP where the quadratic term is in the objective and the constraints are linear where Q is symmetric positive definite. $$ \begin{align*} \min \quad & \frac{1}{2} x^T Q x + ...
0 votes
0 answers
54 views
Could you review my algorithm to assess its conceptual and practical validity for maintaining primal feasibility?
Problem Formulation We aim to solve the following convex quadratic optimization problem: Objective Function: Minimize: $$ f(x) = \frac{1}{2} x^\top Q x + q^\top x, $$ where: ( Q ) is a positive ...
1 vote
0 answers
61 views
Compute hessian for a quadratic cost after nullspace projection
I have a set of quadratic costs, where cost $i$ is of the form $$ f_i(x) = \frac{1}{2} x^T Q_i x + b_ix$$ which means the gradient is $$ \frac{df}{dx}(x) = \sum_i Q_ix + b_i$$. When optimizing these ...
2 votes
0 answers
132 views
Find largest ellipsoid of given axes contained in a box
I want to find the biggest ellipsoid of fixed axes contained within a given box. More precisely, assume I want to maximize the Mahalanobis distance $(x-\mu)^T \Sigma^{-1} (x-\mu)$ where $\mu$ and $\...
1 vote
1 answer
103 views
Closed-form solution to linear regression under positive semi-definite constraint
Let $X,Y\in \mathbb{R}^{d\times n}$ and $W\in\mathbb{R}^{d\times d}$. Is there a closed-form solution to the following minimization problem? $$\min_W \|WX - Y\|_F^2, \quad W \succeq 0.$$
2 votes
1 answer
116 views
solving trace minimization problem in quadratic programming
I'd like to minimize the trace of $P - XHP - PH^T X^T + XSX^T$ with the constraints that every entry of $X \geq 0$ here $P$ is $n$ by $n$ positive definite, $H$ is $m$ by $n$, S is $m$ by $m$ positive ...