gvegayon / abcoptim Goto Github PK

Richard Hodgett writes:

Hi George,

Thank you so much for creating the ABCoptim R package. I am using it to demonstrate ABC with my MSc students.

I have run into a problem which I hope you can help me with. I am trying to use abc_optim() with problems from the globalOptTests package.

Some of the problems in globalOptTests package throw errors:

abc_optim(fn=goTest, fnName="Camel6", par=c(0,0))
Error in if (fun >= 0) return(1/(fun + 1)) else return(1 + abs(fun)) :
missing value where TRUE/FALSE needed
abc_optim(fn=goTest, fnName="Camel3", par=c(0,0))
Error in if (fun >= 0) return(1/(fun + 1)) else return(1 + abs(fun)) :
missing value where TRUE/FALSE needed
abc_optim(fn=goTest, fnName="Powell", par=c(0,0,0,0))
Error in if (r < prob[i]) { : missing value where TRUE/FALSE needed
abc_optim(fn=goTest, fnName="Wood", par=c(0,0,0,0))
Error in if (fun >= 0) return(1/(fun + 1)) else return(1 + abs(fun)) :
missing value where TRUE/FALSE needed

Whilst some of the problems work fine:

abc_optim(fn=goTest, fnName="Ackleys", par=c(0,0,0,0,0,0,0,0,0,0))
$par
[1] -6.623080e+297 -6.623080e+297 -6.623080e+297 -5.879277e+297 -6.623080e+297 -6.623080e+297 -6.623080e+297 -6.623080e+297
[9] -6.623080e+297 -6.623080e+297

$value
[1] 20.00523

$counts
function
53

Is there something I need to define to make the abc_optim() work with the problems that throw the “missing value where TRUE/FALSE needed” error?

Many thanks for any help you can provide,
Kind regards,
Richard

Dr. Richard Edgar Hodgett
Lecturer in Business Analytics and Decision Science