Five variations on a theme by Gyula Farkas (Q1197393): Difference between revisions

Let \(L\) be a linear subspace of \(\mathbb{R}^{n+1}\) and let \(L^ \perp:=\{(y_ 0,\dots,y_ n)\in\mathbb{R}^{n+1}\mid x_ 0y_ 0+\dots +x_ ny_ n=0\) for all \((x_ 0,\dots,x_ n)\in L\}\) denote its orthogonal complement. Then, according to the famous Farkas lemma there exists a vector \({\mathbf a}=(1,a_ 1,\dots,a_ n)\in L\) with \(a_ i\geq 0\) for all \(i=1,\dots,n\) if and only if for every vector \({\mathbf b}=(b_ 0,b_ 1,\dots,b_ n)\in L^ \perp\) with \(b_ 0=1\) there exists some \(i\in\{1,\dots,n\}\) with \(b_ i<0\). The authors consider the case if \(\mathbb{R}\) is replaced by any ordered field \(K\). Hence, all the coefficients \(a_ 1,\dots,a_ n\) of some vector \({\mathbf a}=(1,a_ 1,\dots,a_ n)\in L\subseteq K^{n+1}\) or the coefficients \(b_ 1,\dots,b_ n\) of some vector \({\mathbf b}=(1,b_ 1,\dots,b_ n)\in L^ \perp\) are required to be contained in one and the same subset \(A\subseteq K\) of \(K\). After that they investigate the situation if this requirement is dropped. It means that for any family of subsets \(A_ 1,\dots,A_ n\), \(B_ 1,\dots,B_ n\) of some field \(K\), all containing \(O\in K\), and for any subspace \(L\subseteq K^{n+1}\) one has either a vector \({\mathbf a}=(1,a_ 1,\dots,a_ n)\in L\) with \(a_ i\in A_ i\) for all \(i=1,\dots,n\) or a vector \({\mathbf b}=(1,b_ 1,\dots,b_ n)\in L^ \perp\) with \(b_ i\in B_ i\) for all \(i=1,\dots,n\). Some further extension of the results relating to matroid theory are indicated as well.