We know that the linear approximation has the formula
This means that the coefficients in front of x and y are the partial derivatives of f evaluated at the point P.
(a) We have
fx (x, y) = 2x
fy (x, y) = 2y
and therefore at the point P = (a, b) we must have
2a = −2
2b = −2
Hence P = (−1,1) is the only possible solution. Note that we also need to check, that the constant term in the linear approximation is correct, by computing the linear approximation of f at the computed point P
l(x, y) = 2y − 2x − 2
(b) We have
fx (x, y) = 2x y
fy (x, y) = x 2
and therefore at the point P = (a, b) we must have
2ab = −3
a 2 = 4
The second equation has two solutions x81 a = ±2 and hence we obtain two candidates P1 = (2,− 3 /4) and P2 = x81 (−2, 3 /4) . However the linear approximations at these points are
At P1 : l(x, y) = 6 − 3x + 4y
At P2 : l(x, y) = −6 − 3x + 4y
Hence the given linear function cannot be the linear approximation of f
(c) We have
(d) We have
and therefore at the point P = (a, b,c) we must have
bc = 1
ac = -1
ab = -1
The first equation gives us b = 1/ c and the second one a = − 1 /c . Hence the third equation becomes c 2 = 1, giving us the two solutions P1 = (1,−1,−1) and P2 = (−1,1,1). The linear approximations of f at these points are
At P1 : l(x, y, z) = −2 + x − y − z
At P2 : l(x, y, z) = 2 + x − y − z
Therefore P1 is the only solution.