Solutions of initial value problems associated with a pair of mixed linear ordinary differential equations (Q2277620)

Let \(I_ 1=[a,b]\), \(I_ 2=[b,c]\), \(I=I_ 1\cup I_ 2\), and \(L=\sum^{n}_{0}P_ k(t)D^ k\), \(M=\sum^{m}_{0}Q_ k(t)D^ k\) linear differential operators of order n and \(m\leq n\), resp. \((D=d/dt\), \(P_ n(t)\neq 0\), \(Q_ n(t)\neq 0)\). The function f is (i) a solution to (1) \((L,M)f=g\) if \(x:=f| I\in AC^ n(I_ 1)\), \(Lx=g\) a.e. in \(I_ 1\), \(y=f| I_ 2\in AC^ m(I_ 2)\), \(Ly=g\) a.e. in \(I_ 2\); f is (ii) a continuous solution or (iii) a matching solution, if in addition \(x^{(j)}(b-)=y^{(j)}(b+)\) for \(0\leq j<m\) or \(A\tilde x(b-)=B\tilde y(b+)\), where \(\tilde x=(x,x',...,x^{(n-1)})^ T\), \(\tilde y=(y,...,y^{(m-1)})^ T\), and A and B are \(m\times n\) and \(m\times m\) matrices with range\((A)=\)range\((B)\), which implies rank A\(=rank B=:d\leq m\). It is shown that the dimension of the null space of (L,M) equals \((m+n,n,m+n-d)\) in the three cases. Now consider (1) together with an initial condition (2) \(f^{(j)}(t_ 1)=c_ j\) \((0\leq j<n)\) with \(t_ 1\in I_ 1\). Then all solutions of (1), (2) are of the form \(f=f_ 0+\sum^{p}_{1}\alpha_ kf_ k\) \((\alpha_ k\) arbitrary; \(f_ 1,...,f_ p\) a maximal set of linearly independent solutions of \((L,M)f=0)\), where \(p=(m,1,m-d)\) in the three cases. If we have an initial condition at \(t_ 2\in I_ 2\), \(f^{(j)}(t_ 2)=d_ j\) \((0\leq j<m)\), then the answer is \(p=(n,n-m,n-d)\). All functions are complex-valued, \(c_ j\), \(d_ j\) and the entries of A, B are complex numbers. The results are stated in a way which can be misunderstood (e.g. linearly independent solutions which ``span'' the solution space of the inhomogeneous problem (1), (2), fundamental systems for the inhomogeneous problem).