Analogue package for SOLVER in excel? Other solution?

What is the analogue package for SOLVER in excel?
I have a function with 3 parameters a, b, c and a data vector DATA, the output is a scalar, I need to find a combiner a, b, c to make the result > minimum ?
Is analogue package for SOLVER in excel? Other solution ?
Any hint ?
Paul

Are you solving a linear programming problem? Smooth nonlinear? Nonsmooth nonlinear?

Smooth nonlinear.
Paul

Some help ?
Paul

Optim.jl and JuMP.jl

2 Likes

I obviously have a special interest in promoting Optim.jl, but you can also look to BlackBoxOptim.jl and NLopt.jl depending on your problem type and setup in general (do you want to pass in generic AbstractArrays? Then NLopt.jl won’t work out of the box, etc)

Big thanks, I am traying :slight_smile: but is no easy to me:/

i have my_function(a,b,c), it return x. How (and which package) to find
a ,b, c for the lowest x ? my_function is non linear .

Paul

W dniu 2018-03-12 o 12:01, Patrick Kofod Mogensen pisze:

You can try Optim or any of the other packages mentioned. Do you know more about your problem? Is it continuous? Differentiable?

In Optim you would do something like the following (the absolute minimal example)

x0 = fill(0.0, 3) # use three 0's as starting values for a,b,c
my_vector_function(x) = my_function(a,b,c)
optimize(my_vector_function, x0)

Thanks It is working !

I have multivariate optimization:

x0 = fill(0.5, 3) (makes x0.1, x0.2 x0.3) Is posible |enter the limit
separatly for each other ? like this : |1 > x0.1 >0 |||1 > x0.2 >0| |||2 > x0.3 >0 or one for all ? | Paul |

W dniu 2018-03-12 o 20:08, Patrick Kofod Mogensen pisze:

Sorry, I don’t understand the question.

I hope it is ok ,
x0 = fill(0.5, 3)
optimize(my_fun,x0)

How to write the bounds? Lik thise is ok ? :

optimize(my_fun,x0.|1 > x0[1] >0 |||1 > x0[2] >0| |||2 > x0[3] >0)

?

Paul

W dniu 2018-03-16 o 13:24, Patrick Kofod Mogensen pisze:

http://julianlsolvers.github.io/Optim.jl/latest/user/minimization/#box-minimization

1 Like

Thanks ,
works, but wihout Newton algoritm
Paul

W dniu 2018-03-16 o 13:58, kristoffer.carlsson pisze: