For all $x<0$, $y>0$, let's define the following function on $[-1,1]$ : $$ f_{x,y}(t) = \frac{-2xy}{((y-x) - t*(x+y))^2} $$ The denominator is never zero, so it is well defined. It is specially designed so that (you may verify if you want) : $$ \int_{t=-1}^1 f_{x,y}(t) dt = 1$$ My question is : Does there exist a finite set of abscissae $\{t_0,\dots,t_n\}\in[-1,1]^n$ and positive weights $\{w_0,\dots,w_n\} \in \mathbb{R}^n$ (independent of $x$ and $y$) such that for all values $x<0$ and $y>0$ we have : $$ 1 = \int_{t=-1}^1 f_{x,y}(t) dt = \sum_{i=1}^n w_i f_{x,y}(t_i)$$
Any tip, partial of full solution is welcome.
Substituting $x=-\alpha y$, we have $f_{x,y}(t) = \frac{2\alpha}{(1+\alpha -(1-\alpha)t)^2}$. The question can now be stated for all $\alpha >0$. This is equivalent to ask for the equality of two rational functions with indeterminate $\alpha$. Its partial fraction decomposition wrt $\alpha$ is : $$ f_{x,y}(t) = \frac{\frac{2}{(1+t)^2}}{\alpha + \frac{1-t}{1+t}} - \frac{2\frac{1-t}{(1+t)^3}}{(\alpha + \frac{1-t}{1+t})^2}$$ We ask if there is a finite set of abscissae and weights such that $$1 = \sum_{i=1}^n w_i f_{x,y}(t_i)$$ But this expression would itself be its partial fraction decomposition wrt $\alpha$. Uniqueness of the decomposition thus implies that such abscissea and weights cannot exist.