Package org.moeaframework.problem.MaF
package org.moeaframework.problem.MaF
The MaF test problem suite for many-objective optimization. These problems, often variants of problems found in
other test suites, are intended to represent real-world scenarios.
The PlatEMO implementation, which is where the authors originally developed this test suite, was referenced while developing this Java implementation. A few discrepancies have been observed between the cited paper and the PlatEMO implementation, including:
- MaF5 - Equation 9 in the paper includes
^4
on each objective function, but missing from PlatEMO - MaF6 - Equation 12 uses
1
in the numerator on the second line, but PlatEMO usesPI
- MaF13 - Equation 19 uses
n
but is undefined, based on PlatEMO this value is the same asD
References:
- Cheng, Li, Tian, Zhang, Yang, Jin, and Yao. "A benchmark test suite for evolutionary many-objective optimization." Complex Intell. Syst., 3:67-81, 2017.
- https://github.com/BIMK/PlatEMO
-
ClassDescriptionThe MaF1 test problem, which is a modified inverted version of the DTLZ1 test problem.The MaF10 test problem, which is identical to the
WFG1
test problem.The MaF11 test problem, which is identical to theWFG2
test problem.The MaF12 test problem, which is identical to theWFG9
test problem.The MaF13 test problem.The MaF15 test problem, which is an inverted version of theLSMOP8
test problem withN_k=2
andN_ns=20
..The MaF2 test problem, also known as the "DTLZ2BZ" problem.The MaF3 test problem, which is a convex version of the DTLZ3 problem.The MaF4 test problem, which is an inverted and badly scaled version of the DTLZ3 problem.The MaF5 test problem, which is a convex and badly scaled version of the DTLZ4 problem.The MaF6 test problem, also known as the "DTLZ5(I,M)" problem withI=2
.The MaF7 test problem, which is identical to theDTLZ7
test problem.The MaF8 test problem.The MaF9 test problem.Problem provider for the MaF test problems.