The hessian H of a n-variable function f is a n by n matrix whose (i,j)-th entry is the second order partial derivative:
            2
           d 
H(i,j) = ------ f(x ,x ,...,x )
         dx dx     1  2      n
           i  j
The hessian is symmetric only if the corresponding partial derivatives are continuous. If it is the case, one can gain insight into the behavior of f at a given point a=(a1,a2,...) by computing the determinant of its hessian. For example, if the determinant of the hessian is negative, then a is a saddle point of f. The hessian was introduced by german mathematician Ludwig Otto Hesse in 1842.

See also Jacobian matrix.