What is the solution to finding the top of a hill using gradients and hessian?

  • Thread starter GreenLRan
  • Start date
  • Tags
    Hessian
In summary, the height of a certain hill can be represented as h(x,y) = 10(2xy-3x^2-4y^2-18x+28y+12), where y is the distance north and x is the distance east of South Hadley. To find the top of the hill, the gradient of h must be set to zero, resulting in two equations with two unknowns. Solving for x and y will give the location of the top of the hill. The height of the hill can be found by plugging the location back into the original equation. To find the slope at a specific point, the gradient can be treated as a vector with magnitude equal to the slope and direction pointing towards
  • #1
GreenLRan
61
0

Homework Statement



The height of a certain hill (in feet) is given by h(x,y) = 10(2xy-3x^2-4y^2-18x+28y+12)
where y is the distance (in miles) north, x the distance east of South Hadley.

a) Where is the top of the hill located?
b) How high is the hill?
c) How steep is the slope (in feet per mile) at a point 1 mile north and one mile east of South Hadley? In what direction is the slope steepest, at that point?

Homework Equations



dh = grad h dot dr = |grad h| |dr| cos (theta)

The Attempt at a Solution



I found the gradient of h = 20(y-3x-9)x + 20(x-4y+14)y
I realize setting the gradient of h to zero will give you a max, min, saddle point, or shoulder. I am pretty lost other than that. Any help? Thanks
 
Physics news on Phys.org
  • #2
Evualuate the hessian at each critical point.
 
  • #3
Actually, he should evaluate the determinant of the Hessian at every critical point..
 
  • #4
The Hessian is the determinant, that's the way i learned in school and it's a substantive. While in the formulation "the hessian matrix", it's an adjective.
 
  • #5
Then my school taught it differently. :smile:
 
  • #6
To get a better feel for the equation, note that it is quadratic in both variables. By completing the square you will be able to rewrite this equation as one that describes an ellipse. The center of this ellipse is the top of the hill.
 
  • #7


Though it is clearly way too late to help, I figured I would give what, to me, is the clearest way to solve this.

a) Since your gradient must be equal to zero for a max/min/saddle/etc. and because x components do not add with y components, both your x and y components must be equal to zero. This gives you two equations and two unknowns and you can solve for x and y.

y=3x+9
x=4y-14

Solving these yields only one answer. Since this is a real life problem, common sense tells you that our hill needs a maximum at dh=0, since the hill does not continue into outer space forever or dive deep into the Earth, and we have only one place to put it; at the solution to the two equations above.

b) Plug the location of your hills maximum back into the original equation.

c) The gradient can be defined as a vector with direction in the direction of greatest change and magnitude equal to the slope of the function. You have your gradient already. Plug your given values of x=1 and y=1 into your gradient and treat it like a vector. Its magnitude is your slope and its direction is the direction of greatest change.

Unless this is a very common problem, I imagine we have the same textbook. All the information I have provided is available on the previous page.
 

FAQ: What is the solution to finding the top of a hill using gradients and hessian?

1. What is a gradient in mathematics?

A gradient in mathematics is a vector that points in the direction of the steepest increase of a function. It is defined as the partial derivatives of a multivariable function with respect to each of its independent variables.

2. How is a gradient used in optimization problems?

In optimization problems, the gradient is used to find the direction of the steepest ascent or descent of a function. This direction can then be used to iteratively update the parameters of a model in order to minimize or maximize the function.

3. What is a Hessian matrix?

A Hessian matrix is a square matrix of second-order partial derivatives of a multivariable function. It provides information about the curvature of the function at a specific point and is used to determine whether that point is a minimum, maximum, or saddle point.

4. How is the Hessian matrix related to the gradient?

The Hessian matrix is used to calculate the gradient of a function. The gradient is the vector of first-order partial derivatives, while the Hessian is the matrix of second-order partial derivatives. The gradient is then obtained by multiplying the Hessian matrix with the vector of first-order partial derivatives.

5. What is the significance of the gradient and Hessian in machine learning?

In machine learning, the gradient and Hessian are used in optimization algorithms to train models and find the optimal values for the model's parameters. They play a critical role in determining the direction and speed of the model's updates, leading to more accurate predictions and better performance.

Similar threads

Replies
1
Views
1K
Replies
15
Views
2K
Replies
11
Views
5K
Replies
2
Views
14K
Replies
6
Views
2K
Replies
3
Views
2K
Replies
9
Views
2K
Replies
2
Views
8K
Back
Top