On a class of nonlinear integral equations arising in neutron transport (Q1110765)

A method of continuation is applied to solve the abstract quadratic equation (1) \(x-\lambda B(x,x)-y=0\), (B(x,x) a bounded bilinear symmetric operator in a Banach space X), starting from a solution \(x_ 0\) corresponding to some \(\lambda =\lambda_ 0\), with \(\lambda_ 0^{-1}\) not an eigenvalue of the linear operator \(B(x_ 0,x)\). Theorems are then given in order to specialize the method to the case \(B(x_ 1,x_ 2)(s)=x_ 1(s)(Lx_ 2)(s)+x_ 2(s)(Lx_ 1)(s),\) where \(x_ 1(s)\), \(x_ 2(s)\in C[0,1]\), \((Lx)(s)=1/2\int^{1}_{0}k(s,t)x(t)dt,\) with \(0\leq k(s,t)\leq 1\) (0\(\leq s,t\leq 1)\) and \(k(s,t)+k(t,s)=1\) and with the additional simplifying assumption that the known term y(s) is identically equal to one. In this case, the continuation can start from the trivial solution \(x_ 0(s)=y(s)\equiv 1\), corresponding to \(\lambda_ 0=0\), and continue until \(\lambda\) is equal to some \(\lambda_ c\), an inverse eigenvalue of \(B(x_ c,x)\), for some \(x_ c.\) Theorem 2 is the key of the paper, in that it shows, essentially, that \(\lambda_ c=1\), for the kernel above (more precisely, Th. 2 states that the Fréchet derivative of the L.H.S. of eq. (1) is an invertible operator for \(\lambda\neq 1)\). It appears, however, although not explicitly stated by the author, that the proof of Th. 2 in the paper is valid only in the important, but particular case \(k(s,t)=s/(s+t)\), well known from neutron transport and radiative transfer. This implies that the method is essentially restricted to the Chandrasekhar quadratic integral equation or, possibly, to a related Volterra-like equation. (Some obscurities of the text, together with improper notations or misprints, may require a considerable effort of reconstruction by the reader).