Hi all,

I need to transform (approximate) this equation x =e= Fc/F where all x, fc & f are variables.

Bounds of variables are x.lo = 0; x.up = 1 ; fc.lo = 0 ; fc.up=100; f.lo =0; f.up = 100;

Any suggestions?

Warms Regards,

Rofice

## Reformulation

### Re: Reformulation

Hi Rofice

Why do you want to approximate it? Does it have to be linear? In that case, taking the logarithm would do the job.

Renger

Why do you want to approximate it? Does it have to be linear? In that case, taking the logarithm would do the job.

Renger

### Re: Reformulation

Hi Renger,

Reason for approximation is, I am making MIP model.

I will be pleased if you can share your thoughts on this, Actually I solved this model completely as MINLP but later my boss want global solution. So my question is why we need global solution? As far as I know local solution may miss true solution thats why global solution is preferred but in MIP modelling if product or division or power of variables is required then one has to approximate solution again. So what is the difference because approximation may also overestimate or underestimate?

Is this possible that global solution can be obtained by nlp or minlp?

Reason for approximation is, I am making MIP model.

I will be pleased if you can share your thoughts on this, Actually I solved this model completely as MINLP but later my boss want global solution. So my question is why we need global solution? As far as I know local solution may miss true solution thats why global solution is preferred but in MIP modelling if product or division or power of variables is required then one has to approximate solution again. So what is the difference because approximation may also overestimate or underestimate?

Is this possible that global solution can be obtained by nlp or minlp?