Saddle Point definition
In multivariable calculus, a saddle point is a stationary point with both higher and lower function values arbitrarily close by.
Saddle Point in a picture
How it works
For f(x, y) = x² − y², the origin has gradient zero. Along y = 0 the values are positive away from the origin, while along x = 0 they are negative. For a twice continuously differentiable function at a critical point, the second derivative test detects a saddle when fₓₓfᵧᵧ − (fₓᵧ)² is negative.
Worked examples
Classify (0, 0) for f = x² − y².
Saddle point. The two axes give changes in opposite directions.
Does a stationary point have to be a local maximum or minimum?
No. It can be a saddle.
If the Hessian determinant is zero at a critical point, what does the standard test decide?
It is inconclusive.
Common mistake
A zero gradient only marks a candidate. Examine nearby values or apply a valid second derivative test.
Quick check
What is the gradient at a stationary saddle?
Show answer
Zero.
Can a saddle point be a local minimum?
Show answer
No. There are lower values arbitrarily nearby.
What does a negative determinant show at a critical point under the standard test?
Show answer
A saddle point.
Sources
Flashcard
Saddle Point
Reveal the definition
In multivariable calculus, a saddle point is a stationary point with both higher and lower function values arbitrarily close by.
Effortless Math. “Saddle Point.” Math Dictionary.
