f(x,y) = e^y(y^2-x^2)
so the sationary points are determined by solving
grad f(x,y) = vec 0
or
{
(-2 e^y x = 0), (2 e^y y + e^y (-x^2 + y^2) = 0)
:}
giving two solutions
((x=0,y=0),(x=0,y=-2))
Those points are qualified using
H = grad(grad f(x,y))
or
H =((-2 e^y, -2 e^y x),(-2 e^y x, 2 e^y + 4 e^y y + e^y (-x^2 + y^2)))
so
H(0,0) = ((-2, 0),(0, 2)) has eigenvalues {-2,2}. This result qualifies point {0,0} as a saddle point.
H(0,-2)=((-2/e^2, 0),(0, -2/e^2)) has eigenvalues {-2/e^2, -2/e^2}. This result qualifies point {0,-2} as a local maximum.
Attached the f(x,y) contour map near the points of interest