The implicit function theorem

Try

The equation x²+y²=4 describes a circle, not globally one function y of x. Near (0,2), however, the upper branch y=√(4−x²) is a smooth function of x. Near (2,0), two nearby y values occur for x just below 2, so a single y graph cannot describe all nearby zeros. The implicit function theorem identifies which variable can be solved for locally.

Predict

Predict the relevant derivative. For F(x,y)=x²+y²−4, solving for y requires F_y=2y to be nonzero at the base point. It succeeds at (0,2) and fails at (2,0). At the latter point, F_x=4 is nonzero, so solving for x as a function of y works instead. The choice of dependent variable is mathematical information, not merely a change of notation.

Experiment

Track the block structure for a general constraint.

Object	Dimensions
Independent variable x	R^n
Solved variable y	R^m
Constraint F(x,y)=0	R^m
D_yF	m×m, required invertible
Dg	m×n

The number of solved coordinates matches the number of equations in this theorem. A general system with inequalities, singular constraints or rank deficiency needs different analysis; the square derivative block is not interchangeable with the full Jacobian.

Understand

Let F be C¹ on an open neighborhood of (a,b), with F(a,b)=0 and invertible D_yF(a,b). Define H(x,y)=(x,F(x,y)). Its derivative has block form [[I,0],[D_xF,D_yF]], so it is invertible exactly when the lower-right block is. The inverse function theorem gives a C¹ local inverse for H.

Restrict that inverse to output points (x,0) near (a,0). Its first coordinate must equal x because H preserves the first coordinate. Write the second coordinate as g(x). Then F(x,g(x))=0 and g(a)=b. Local injectivity of H ensures these are all nearby zeros and gives uniqueness among nearby y values. Open neighborhoods can be reduced to a product neighborhood so this description applies for every x in one fixed small open set.

Apply

Differentiate F(x,g(x))=0 using the chain rule: D_xF(x,g(x))+D_yF(x,g(x))Dg(x)=0. The y-block remains invertible after shrinking the neighborhood, giving Dg=−(D_yF)⁻¹D_xF along the graph. For the circle at (0,2), the derivative is zero; at a nearby upper-branch point it is −x/y, matching direct differentiation.

Repair “if the y derivative vanishes, there is no implicit function.” The theorem gives a sufficient condition. The equation y³=0 has unique smooth solution y=0 despite a zero y derivative along every solution. In contrast, y²−x²=0 has two crossing branches through the origin. Failure of the sufficient hypothesis permits several outcomes and must be analyzed directly.

Continue

Optional depth interprets the derivative formula as local sensitivity: small independent-variable changes produce approximately Dg times that change in the solved variables. Near a nearly singular y-block, the inverse matrix can be large, warning that small input perturbations may require large output changes. Local existence alone is not a numerical conditioning guarantee.

Finish a proof portfolio containing a uniform-limit argument, a metric compactness equivalence, a contraction construction and a local calculus theorem with an explicit failed-hypothesis example. This sequence develops Analysis II foundations; measure theory, infinite-dimensional calculus and independent course-level assessment remain distinct further work.