Skip to content

Latest commit

 

History

History
32 lines (20 loc) · 2.04 KB

README.md

File metadata and controls

32 lines (20 loc) · 2.04 KB

MI-PAA Course

Problémy a algoritmy

CTU / ČVUT FIT

Task 1 - naive and aproximative Knapsack solver

We were implementing a naive and slightly improved aproximative solution using heuristic for the Knapsack problem. The main goal was to measure error of the approximative solution. The results contains graphs and the final report.

Solution can be compiled using GNU make on Linux and Visual Studio on Windows.

Task 2 - dynamic decomposition and FPTAS Knapsack solver

In the second part we took an advantage of previous one - another improvement was implementation of branch and bounds prunning. This approach effectively removes states which would never lead to a better solution that the current.

Another approach we implemented was dynamic decomposition and fast polynomial-time approximation scheme (FPTAS).

This version can be compiler only using the GNU make.

Task 3 - eomparison of methods

We took previously programmed solver on the Knapsack problem and let it crush a lot of different instances. The basic difference was that results depend on the instance properties.

Folder contains the final report as well as graphs and exact runtimes.

Task 4 - evolution algorithm Knapsack solver

The last algorithm trying to solve the Knapsack problem was evolutin algorithm. As this is random algorithm results may (and will) change over time.

The solution was created for the Visual Studio

Semestral work - evolution algorithm 3SAT solver

Our semestral work was focused on 3SAT problem. We took an advantage of previous experience with evolution programming. I was enough to copy solution of the 4th task and bend it. I had changed problem representation in the genom and added another parameter tuning features. My approach was to use parameter grid, when we iterate over each parameter's value until all parameter space is exhausted.

The final report describing my approach, overall quality and results is in the place.

Unfortunately this approach was not fully appreciated during the evaluation. You have been warned!