LP Formulation of - and -norm Regression
The -norm regression problem:
where , are given, can be written as the LP
where stands for the -th row of .
Similarly, the -norm regression problem:
can be written as the LP
CVX does not require the user to transform the problem into LP format. The following syntax will work for any :
CVX implementation
cvx_begin
variable x(n,1)
minimize( norm(A*x-y,p) )
cvx_end
|