5.9. pyopus.problems.karmitsa — Large scale nonsmooth test functions (Karmitsa set)

Large scale nonsmooth test functions (Karmitsa set) (PyOPUS subsystem name: KARMITSA)

This module is independent of PyOPUS, meaning that it can be taken as is and used as a module in some other package. It depends only on the cpi and the _karmitsa modules.

[karmitsa]

Karmitsa N.: Test Problems for Large-Scale Nonsmooth Minimization, Technical report B.4/2007, University of Jyvaskyla, Jyvaskyla, 2007.

class pyopus.problems.karmitsa.LSNSU(name=None, number=None, n=2)

Unconstrained problems from the Karmitsa test suite. Problems 10 and 11 were added by Á. Bűrmen.

• name - problem name
• number - problem number (0-11)
• n - problem dimension

Attributes:

• name - problem name
• n - number of variables
• initial - initial values of variables
• fmin - best known minimum

This module is independent of PyOPUS, meaning that it can be taken as is and used as a module in some other package. It depends only on the cpi and the _karmitsa modules.

cpi()

Returns the common problem interface.

Subgradient is not supported.

xmin is also not available.

See the CPI class for more information.

f(x)

Returns the value of the function at x.

names = ['GeneralizedMAXQ', 'GeneralizedMXHILB', 'ChainedLQ', 'ChainedCB3I', 'ChainedCB3II', 'ActiveFaces', 'GeneralizedBrown2', 'ChainedMifflin2', 'ChainedCrescentI', 'ChainedCrescentII', 'GeneralizedL1HILB', 'GeneralizedWatson']

List of all function names

class pyopus.problems.karmitsa.LSNSB(name=None, number=None, n=2, feasible=1)

Bound constrained problems from the Karmitsa test suite.

• name - problem name
• number - problem number (0-9)
• n - problem dimension
• feasible - 0=initial point unchanged, 1=feasible, 2=strictly feasible, default=1

Attributes:

• name - problem name
• n - number of variables
• xl - lower bounds on variables
• xh - upper bounds on variables
• initial - initial values of variables
• fmin - best known minimum

This module is independent of PyOPUS, meaning that it can be taken as is and used as a module in some other package. It depends only on the cpi and the _karmitsa modules.

cpi()

Returns the common problem interface.

Subgradient is not supported.

xmin is also not available.

See the CPI class for more information.

f(x)

Returns the value of the function at x.

names = ['GeneralizedMAXQ', 'GeneralizedMXHILB', 'ChainedLQ', 'ChainedCB3I', 'ChainedCB3II', 'ActiveFaces', 'GeneralizedBrown2', 'ChainedMifflin2', 'ChainedCrescentI', 'ChainedCrescentII']

List of all function names

class pyopus.problems.karmitsa.LSNSI(name=None, number=None, cname=None, cnumber=None, n=10)

Inequality constrained problems from the Karmitsa test suite.

• name - function name
• number - function number (0-9)
• cname - constraint name
• cnumber - constraint number (0-7)
• n - problem dimension

Attributes:

• name - problem name
• fname - function name
• cname - constraint name
• n - number of variables
• m - number of constraints
• cl - lower bounds on constraint functions
• ch - upper bounds on constraint functions
• initial - initial values of variables

This module is independent of PyOPUS, meaning that it can be taken as is and used as a module in some other package. It depends only on the cpi and the _karmitsa modules.

c(x)

Returns the value of the constraints at x.

cnames = ['TridiagonalI', 'TridiagonalII', 'MAD1I', 'MAD1II', 'ModifiedMAD1I', 'ModifiedMAD1II', 'P20I', 'P20II']

List of all constraint names

cpi()

Returns the common problem interface.

Subgradient is not supported.

xmin is also not available.

See the CPI class for more information.

f(x)

Returns the value of the function at x.

names = ['GeneralizedMAXQ', 'GeneralizedMXHILB', 'ChainedLQ', 'ChainedCB3I', 'ChainedCB3II', 'ActiveFaces', 'GeneralizedBrown2', 'ChainedMifflin2', 'ChainedCrescentI', 'ChainedCrescentII']

List of all function names

Example file karmitsa.py in folder demo/problems/

# Karmitsa test suite

from pyopus.problems.karmitsa import *

if __name__=='__main__':
	print("Unconstrained, n=50")
	for ii in range(10):
		prob=LSNSU(number=ii, n=50)
		print("%d: %25s: f0=%e" % (ii, prob.name, prob.f(prob.initial)))
	print()
	
	print("Bound constrained, n=50")
	for ii in range(10):
		prob=LSNSB(number=ii, n=50)
		print("%d: %25s: f0=%e" % (ii, prob.name, prob.f(prob.initial)))
	print()
	
	print("Inequality constrained, n=10")
	for ii in range(10):
		for jj in range(8):
			prob=LSNSI(number=ii, cnumber=jj, n=10)
			c=prob.c(prob.initial)
			feas=((c>prob.cl) & (c<prob.ch)).all()
			print("%d, %d: %30s: f0=%12.4e feas=%d" % (ii,jj, prob.name, prob.f(prob.initial), feas))
	print()