Improvement heuristics I - Local Search, steepest descent

17 important questions on Improvement heuristics I - Local Search, steepest descent

What is the definition of a local search alogrithm?

An algorithm that constructs a final solution by starting from some initial  (complete) feasible solution and iteratively modifying the current solution to obtain a new one until a better solution is obtained in this manner.
  • Start with feasible solution;
  • Repeatedly search neighborhood for improvements; 

Give [4] steps of a local search algorithm;

  1. Start with given initial solution;
  2. Search neighborhood for a better solution using some operation;
  3. Accept or reject the solution and start again;
  4. Stop if local optimum or stopping criteria is reached;

Give the definition of local search, steepest descent

Steepest descent method searches the entire neighbourhood for the best neighbour based on some operator and accepts if it does not decrease the objective value.

The method stops when a local optimum is reached.
  • Higher grades + faster learning
  • Never study anything twice
  • 100% sure, 100% understanding
Discover Study Smart

Explain the neighborhood operator k-OPT

K-opt is a local search operator where in every swap k exchanges take place.

When is a tour k-optimal?

When there are no further improvements to be made using the k operator.

What is the idea of OR-opt?

Number of exchanges can become very large for a large k, or-opt only considers only a part of the exchange.

Steepest descent
  • [1] advantage
  • [2] disadvantages

Advantage
  • Quickly generates a solution


Disadvantage
  • Converges fast
  • Get stuck easily in a local optimum

How can local optimality be overcome? Also, give the name of these.

Heuristics that balance intensification and diversification.

Metaheuristics   

How does GRASP apply the concept of intensification and diversification?

Intensification through local search. Diversification through adaptive search.

What is a method to overcome cycling?

Tabu search

What are possible attributes for a recently visited solution that is stored in the tabu-list?

  • Entire solution;
  • Moves - operators;
  • Changed items; 

What is meant with aspiration level in tabu search?

Rule that allows changes considering a solution although these are in the tabu-list.

Why is an aspiration level sometimes necessary?

  • Global best | Best overall solution is found;
  • Region best | Best solution with a certain property is found;
  • Recent best | Best recently visited solution;

Give the [5] tabu search implementation decisions

  1. Initial solution;
  2. Neighbourhood operator;
  3. Tabu list length;
  4. Tabu list attributes;
  5. Stopping criteria;

What is meant with connectivity, Why is it needed?

From every possible solution all other solutions can be reached by repeatedly applying the chosen operator.

Needed to reach global optimum.

What influence has the tabu list length on a solution? How can static and dynamic tabu lists be of use?

Dynamic lists can increase in length if the best solution is not updated, and decrease if a certain time has passed without solution revisits.

Give [2] stopping criteria considering the tabu list

  • If improvements are not realized after a predetermined number of exchanges;
  • If all the direct neighbours are in the tabu list;

The question on the page originate from the summary of the following study material:

  • A unique study and practice tool
  • Never study anything twice again
  • Get the grades you hope for
  • 100% sure, 100% understanding
Remember faster, study better. Scientifically proven.
Trustpilot Logo