Minimization Alignment

Name:

Restore previous session, if applicable
Start blank quiz
Note:

If you wish to ask for a hint or a solution, or if you want your answers checked, you have to log in. Your answers, or the fact that you asked for a hint or solution, may be recorded.




Usage notes


For each of the following functions and for each minimization method described in class, determine which of the following cases applies:
  • method is not applicable
  • method is applicable and recommended
  • method is applicable, but not recommended
  • method converges to the minimum in one iteration

and choose which of the following comments (referenced as "Comment 1", ..., "Comment 7" in the table below) applies best for each function.

Note: Inverse parabolic interpolation was not covered by Prof. Gonnet this year, but you can find it in section 1.5.2.2 of the online script.
  1. The method is applicable to one-dimensional problems only.
  2. Brent includes Golden section search, but converges much faster for smooth functions.
  3. Non-smooth function makes inverse parabolic interpolation of Brent's method useless.
  4. For functions in several variables, random direction search is better than steepest descent or Newton only in case of non-smooth functions. All of f1, f4 and f5 are smooth enough.
  5. Random directions and steepest descent are not applicable per se to one-dimensional problems, only the line search algorithm used in both can be aplied to one-dimensional problems.
  6. Newton-Raphson converges in one step to the minimum when applied to quadratic functions.
  7. The scalar Newton-Raphson is applicable (and quickly convergent) for functions smooth enough. The function f3 is not differentiable.



Function Method: case and comment
f3(x1) = x1 - |x1-5| + |3x1 + 10|
 
1. Golden section search

2. Brent's method

3. Random directions

4. Steepest descent and conjugate gradients

5. Newton and spectral method

f4(x1, x2) = |x1 - 3|2 + |x2 + 1|2 - x1x2 + 3x1 + 2
 
1. Golden section search

2. Brent's method

3. Random directions

4. Steepest descent and conjugate gradients

5. Newton and spectral method

f5(x1, x2) = (x1 - 3x2) / (x12 + x22 + 10)
 
1. Golden section search

2. Brent's method

3. Random directions

4. Steepest descent and conjugate gradients

5. Newton and spectral method




Home    

Feedback to quiz developers, problem reports...please use it!!!


Powered by PearlQuiz. With assistance from SkillsOnline and Web Pearls.