Caution: 0-dimensional slice

[renan]

Hello Mr. Fourier

I'm glad to join the group...

could you please give me an idea why im getting this caution...

any help would mean a lot.

----------------------------------------------------------------------------

set PRODUCT   ;

set CROSS := {i in PRODUCT, j in PRODUCT} ; 

param phi {CROSS} ;

maximize PROFIT:

sum {i in PRODUCT}alpha[i]*width[i]^beta[i]*

prod{(i,j) in CROSS} width[j]^phi[i,j]*

(

(selling[i]-unit[i]-replenish[i])*choose[i]*(facings[i])^beta[i]* 

prod{(i,j) in CROSS} facings[j]^phi[i,j]-

(restock[i]/hold[i])*choose[i]*(facings[i])^beta[i]-1*

prod{(i,j) in CROSS} facings[j]^phi[i,j]

)

- sum {i in PRODUCT} ((holding[i]*choose[i]*facings[i])-fixed*choose[i])

irion.txt, line 39 (offset 1031):

Caution: 0-dimensional slice

context:  prod{(i,j) in  >>> CROSS} <<<  facings[j]^phi[i,j]-

irion.txt, line 42 (offset 1133):

Caution: 0-dimensional slice

context:  prod {(i,j) in  >>> CROSS} <<<  facings[j]^phi[i,j]

ampl: 

[renan]

Apologies, Mr. Fourer i mean...

ive been trying all possibilities on this for hours

could you shed some light please...many thanks

[Robert Fourer]

Your 2nd and 3rd "prod {(i,j) in CROSS}" are part of what is being multiplied by the first "prod {(i,j) in CROSS}". Thus you are trying to index over {(i,j) in CROSS} when both i and j are already defined, which is the reason for the "0-dimensional slice" warning. To keep the first product from extending to the others, you can parenthesize it:

maximize PROFIT:
sum {i in PRODUCT} alpha[i] * width[i]^beta[i] *

(prod {(i,j) in CROSS} width[j]^phi[i,j]) * (...

If not parenthesized in this way, sums and products apply to everything until the next + sign.

Bob Fourer
am...@googlegroups.com

=======

[renan maglasang]

Dear Mr. Fourer,

I thank you very much....its solved. its solved.

--
You received this message because you are subscribed to a topic in the Google Groups "AMPL Modeling Language" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ampl/aKsh2MUKE4o/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ampl+uns...@googlegroups.com.
To post to this group, send email to am...@googlegroups.com.
Visit this group at http://groups.google.com/group/ampl.
For more options, visit https://groups.google.com/d/optout.

[renan maglasang]

Hello Mr. Fourer

again...any help on this...

i want that if choose=0, facings would be equals to 1...

(this is to avoid my objective function from being zero since its involves crossproduct)

i believe the constraint below would make this way possible but my question is why AMPL gives infeasible solution?

var facings {i in PRODUCT} >=0 integer; 

var choose {i in PRODUCT} >=0 binary ;

subject to POSITIVEINT{i in PRODUCT}:

(width[i]*facings[i]-width[i])*(width[i]*choose[i]+width[i]) >= 0;

[renan maglasang]

​oops correction for my constraints...should be all minus width...

however, still it didnt work...

im not sure but is setting:​

var choose {i in PRODUCT} >=1 binary ;

a valid circumvention to my original intention?

[Robert Fourer]

It is best to implement logic with linear rather than quadratic constraints. You might write

facings[i] >= 1 - choose[i]

to force facings[i] >= 1 when choose[i] = 0. If in fact you need to have facings[i] = 1 when choose[i] = 0, then you could also specify a constraint like

facings[i] <= 1 + U[i]*choose[i]

where U is an upper bound on the value of facings[i]-1. (When choose[i] = 1, these constraints have no effect.)

Bob Fourer
am...@googlegroups.com

=======