I am faced with the following system of equations and I'm looking for tools that allow me to characterize its solutions. The unknowns are $\{x_i\}$, $\{y_i\}$, and $\lambda$, for $i=1,...,N$. So there are $2N+1$ equations and the same number of unknowns. Note the subscript for the summations. The $\sigma_{ij}$ are entries of a full-rank, invertible $N$ by $N$ matrix. If helpful, I know the following properties of the parameters: $$\sum_i \beta_i = 1$$ $$\sum_j \sigma_{ij} = 1-\alpha_i > 0$$ The system of equations I want to solve is the following: $$x_i +\alpha_i \lambda = \sum_j \sigma_{ij} y_j$$ $$\frac{1}{x_j} - \beta_j = \sum_j \sigma_{ij} \frac{1}{y_j}$$ $$\sum_i \frac{1}{x_i} = \sum_i \frac{1}{y_i}$$