Check the critical point first
The Hessian collects the second partial derivatives of a scalar function. For a twice continuously differentiable function, mixed partials agree and the matrix is symmetric. The local extremum test needs more than this matrix: the point must lie inside the domain and have zero gradient. Only then does curvature classify the stationary candidate.[1]
Evaluate every derivative at the same previously checked point.[1]
If D>0 and fxx>0, there is a strict local minimum; if D>0 and fxx<0, a strict local maximum. If D<0 there is a saddle point. If D=0 the test is inconclusive. A local test does not solve a global optimization problem or replace checking the boundary of a restricted domain. The gradient and stationary points must therefore be examined before applying this test.[1]
Why the determinant distinguishes directions
At a stationary point the linear Taylor term vanishes. A definite Hessian makes the quadratic form determine the local sign of change.[1]
In two dimensions, D<0 means the eigenvalues have opposite signs, producing rising and falling directions. When D>0 the signs agree, and fxx distinguishes positive from negative definiteness. D=0 leaves at least one second-order direction unresolved, so higher-order terms may matter. For three or more variables the determinant alone does not provide the same complete test. The Taylor approximation explains why this quadratic change dominates locally when its sign is definite.[1]
Distinguish a minimum from a saddle
Positive definite quadratic form
- Given
- f=x2+xy+2y2.
- Find
- Classify the origin.
The gradient is (2x+y,x+4y), vanishing only at (0,0).
Find the second derivatives and determinant.
D>0 and fxx=2>0 give a strict local minimum.
Result: The origin is a strict minimum.
Check: Completing the square additionally proves global nonnegativity with equality only at the origin. That global claim follows from this particular function, not from every Hessian test.
A mixed term produces a saddle
- Given
- f=x2+4xy+y2.
- Find
- Check the stationary origin.
The gradient (2x+4y,4x+2y) vanishes at the origin.
The determinant is negative.
Two lines show the opposite signs directly.
Result: The origin is a saddle point.
Check: Both pure second derivatives are positive, but the mixed term prevents positive definiteness.
Recall the method later with the formula’s flashcards.
10 flashcards · English materials
Already have an account? Sign in for flashcardsAvailable from Essential. A ready-made formula set, no AI credits required. Article exercises are not all included.
Calculator for a symmetric 2×2 Hessian
The specialized calculator accepts a=fxx, b=fxᵧ and c=fᵧᵧ as already established values and computes D=ac−b2. Its classification notes apply only after you have checked stationarity, an interior point and the required C2 regularity. It does not differentiate arbitrary functions or infer those assumptions from three numbers. At D=0 it does not assert a minimum, maximum or saddle.[1]
Practise assumptions and degenerate cases
Calculate · Foundation
Classify the origin for f=−2x2−y2.
Show hint
Calculate the gradient first.
Learning goal: Check the sign as well as the determinant.
Show solution and calculation
∇f=(−4x,−2y), so the origin is stationary.
H=diag(−4,−2), D=8>0 and fxx<0: strict local maximum.
Interpret · Application
For f=x4+y4, H(0,0)=0. Does this mean no extremum?
Show hint
Compare function values directly.
Learning goal: Distinguish inconclusive from disproven.
Show solution and calculation
No.
D=0 only makes the Hessian test inconclusive.
Since x4+y4≥0 with equality only at (0,0), the origin is a strict minimum.
Error analysis · Foundation
Someone calls (0,0) a minimum of f=x2+y2+x because H=2I. What is missing?
Show hint
Check fx at the origin.
Learning goal: Check stationarity before curvature.
Show solution and calculation
∇f(0,0)=(1,0)≠0.
The origin is not stationary, so this extremum test cannot be applied there.
The minimum is at (−1/2,0).
Calculate · Application
At a stationary C2 point, fxx=3, fxᵧ=2 and fᵧᵧ=5. Calculate D and classify.
Show hint
Square the mixed derivative.
Learning goal: Square the mixed term correctly.
Show solution and calculation
D=3·5−22=11>0; fxx>0, giving a strict local minimum.
Transfer · Challenge
A symmetric 3×3 matrix has eigenvalues −1,−1,1 and determinant 1. Does the positive determinant prove a minimum?
Show hint
Positive definiteness requires all eigenvalues to be positive.
Learning goal: Respect the test’s dimensional scope.
Show solution and calculation
No.
Two negative eigenvalues and one positive eigenvalue make the quadratic form indefinite.
The two-dimensional determinant rule cannot be generalized on its own.
Check your understanding with the formula’s reference test.
8 test questions · English materials
Already have an account? Sign in for the testAvailable from Essential. A ready-made formula set, no AI credits required. Article exercises are not all included.
Sources for this explanation
Quanta · Updated November 1, 2026
Editorial responsibility: Quanta Study editorial team
Review scope
This explanation, 2 worked examples and 5 exercises are available in German and English. Structure, source references and mathematical notation undergo automated checks. Calculations, model assumptions and both language versions receive a separate AI review.
An independent, named human subject review has not yet been documented. Our exercise numbers are worked examples, rather than measurements published by the cited sources.
Editorial responsibility, sources and methodsChange history
- Authored a distinct DE/EN article with model limits, two worked examples and five varied exercises.
Cite this reference page
Quanta. Hessian matrix and second-derivative test. Updated November 1, 2026. https://quanta-study.de/en/formula-sheet/hesse-matrix-zweiter-ableitungstest
When citing a specific scientific statement, also cite the specialist source linked to that statement.