I am using Gurobi 7.0 through Matlab. Based on the documentation, in order to find the n best solutions you need to set the parameters:
PoolSearchMode=2, to find alternative optimal solutions in a systematic way.
PoolSolutions=n, number of of solution in the pool.
When I do this my result contains the same fields as with the default parameters, i.e. only one solution. I have also tried changing the parameter SolutionNumber, but it does not affect the outcome.
I suspect the alternative optimal solutions are being found, since the solver reports on the prompt a solution count equivalent to n with objective values, but I am not able to retrieve them. I hope this is not another limitation of the Gurobi Matlab API.
Also, I know I could find these solutions using integer cuts, but from my understanding that would be much more inefficient since it would require to start the branch and bound tree from the beginning.
It is not possible. The Gurobi Matlab interface is limited because it does not treat the model as a class, even though Matlab offers object oriented programming. This limits many functionalities. CPLEX however allows Matlab users to interact with the model class and retrieve solutions from the solutions pool.
Related
I was wondering if there exists a technical way to choose initial parameters to these kind of problems (as they can take virtually any form). My question arises from the fact that my solution depends a little on initial parameters (as usual). My fit consists of 10 parameters and approximately 5120 data points (x,y,z) and has non linear constraints. I have been doing this by brute force, that is, trying parameters randomly and trying to observe a pattern but it led me nowhere.
I also have tried using MATLAB's Genetic Algorithm (to find a global optimum) but with no success as it seems my function has a ton of local minima.
For the purpose of my problem, I need justfy in some manner the reasons behind choosing initial parameters.
Without any insight on the model and likely values of the parameters, the search space is too large for anything feasible. Think that just trying ten values for every parameter corresponds to ten billion combinations.
There is no magical black box.
You can try Bayesian Optimization to find a global optimum for expensive black box functions. Matlab describes it's implementation [bayesopt][2] as
Select optimal machine learning hyperparameters using Bayesian optimization
but you can use it to optimize any function. Bayesian Optimization works by updating a prior belief over a distribution of functions with the observed data.
To speed up the optimization I would recommend adding your existing data via the InitialX and InitialObjective input arguments.
I am very new to MatLab. Thus I am sorry if this is very basic.
I use a function called fmincon to do find a solution for minimizing a function. Why do I get different solutions for running fmincon?
I would like to know a satisfying or convincing mathematical or programming explanation for having different solutions using fmincon.
Check these limitations in the MATLAB documentation.
fmincon is a gradient-based method that is designed to work on problems where the objective and constraint functions are both continuous and have continuous first derivatives.
The function is very delicate and it is best if you can avoid it. It only works neatly on problems that are neatly defined to begin with. Any deviation can lead to local instead of global minima, and these can depend (among other things) on your initial solution estimate or starting point.
As fmincon is sensitive to initial point, If you set different start point for the fmincon, you might get a different solution in each apply. You can find one of the algorithms of fmincon here.
When running the GlobalSearch solver on a nonlinear constrained optimization problem I have, I often get very different solutions each run. For the cases that I have an analytical solution, the numerical results are less dispersed than the non-analytical cases but are still different each run. It would be nice to get the same results at least for these analytical cases so that I know the optimization routine is working properly. Is there a good explanation of this in the Global Optimization Toolbox User Guide that I missed?
Also, why does GlobalSearch use a different number of local solver runs each run?
Thanks!
A full description of how the GlobalSearch algorithm works can be found Here.
In summary the GlobalSearch method iteratively performs convex optimization. Basically it starts out by using fmincon to search for a local minimum near the initial conditions you have provided. Then a bunch of "trial points", based on how good the initial result was, are generated using the "scatter search algorithm." Then there is some more convex optimization and rating of "how good" the minima around these points are.
There are a couple of things that can cause the algorithm give you different answers:
1. Changing the initial conditions you give it
2. The scatter search algorithm itself
The fact that you are getting different answers each time likely means that your function is highly non-convex. The best thing that I know of that you can do in this scenario is just to try the optimization algorithm at several different initial conditions and see what result you get back the most frequently.
It also looks like there is something called the 'PlotFcns' property which would allow you get a better idea what the functions the solver is generating for you look like.
You can use the ga or gamulti objective functions within the GlobalSearch api. I would recommend this. Convex optimizers wont be able to solve a non-linear problem. Even then Genetic Algorithms dont gaurantee the solution. If you run the ga and then use its final minimum as the start of your fmincon search then it should result in the same answer consistently. There may be better ones but if the search space is unknown you may never know.
Are there any faster and more efficient solvers other than fmincon? I'm using fmincon for a specific problem and I run out of memory for modest sized vector variable. I don't have any supercomputers or cloud computing options at my disposal, either. I know that any alternate solution will still run out of memory but I'm just trying to see where the problem is.
P.S. I don't want a solution that would change the way I'm approaching the actual problem. I know convex optimization is the way to go and I have already done enough work to get up until here.
P.P.S I saw the other question regarding the open source alternatives. That's not what I'm looking for. I'm looking for more efficient ones, if someone faced the same problem adn shifted to a better solver.
Hmmm...
Without further information, I'd guess that fmincon runs out of memory because it needs the Hessian (which, given that your decision variable is 10^4, will be 10^4 x numel(f(x1,x2,x3,....)) large).
It also takes a lot of time to determine the values of the Hessian, because fmincon normally uses finite differences for that if you don't specify derivatives explicitly.
There's a couple of things you can do to speed things up here.
If you know beforehand that there will be a lot of zeros in your Hessian, you can pass sparsity patterns of the Hessian matrix via HessPattern. This saves a lot of memory and computation time.
If it is fairly easy to come up with explicit formulae for the Hessian of your objective function, create a function that computes the Hessian and pass it on to fmincon via the HessFcn option in optimset.
The same holds for the gradients. The GradConstr (for your non-linear constraint functions) and/or GradObj (for your objective function) apply here.
There's probably a few options I forgot here, that could also help you. Just go through all the options in the optimization toolbox' optimset and see if they could help you.
If all this doesn't help, you'll really have to switch optimizers. Given that fmincon is the pride and joy of MATLAB's optimization toolbox, there really isn't anything much better readily available, and you'll have to search elsewhere.
TOMLAB is a very good commercial solution for MATLAB. If you don't mind going to C or C++...There's SNOPT (which is what TOMLAB/SNOPT is based on). And there's a bunch of things you could try in the GSL (although I haven't seen anything quite as advanced as SNOPT in there...).
I don't know on what version of MATLAB you have, but I know for a fact that in R2009b (and possibly also later), fmincon has a few real weaknesses for certain types of problems. I know this very well, because I once lost a very prestigious competition (the GTOC) because of it. Our approach turned out to be exactly the same as that of the winners, except that they had access to SNOPT which made their few-million variable optimization problem converge in a couple of iterations, whereas fmincon could not be brought to converge at all, whatever we tried (and trust me, WE TRIED). To this day I still don't know exactly why this happens, but I verified it myself when I had access to SNOPT. Once, when I have an infinite amount of time, I'll find this out and report this to the MathWorks. But until then...I lost a bit of trust in fmincon :)
I have to solve a multiobjective problem but I don't know if I should use CPLEX or Matlab. Can you explain the advantage and disadvantage of both tools.
Thank you very much!
This is really a question about choosing the most suitable modeling approach in the presence of multiple objectives, rather than deciding between CPLEX or MATLAB.
Multi-criteria Decision making is a whole sub-field in itself. Take a look at: http://en.wikipedia.org/wiki/Multi-objective_optimization.
Once you have decided on the approach and formulated your problem (either by collapsing your multiple objectives into a weighted one, or as series of linear programs) either tool will do the job for you.
Since you are familiar with MATLAB, you can start by using it to solve a series of linear programs (a goal programming approach). This page by Mathworks has a few examples with step-by-step details: http://www.mathworks.com/discovery/multiobjective-optimization.html to get you started.
Probably this question is not a matter of your current concern. However my answer is rather universal, so let me post it here.
If solving a multiobjective problem means deriving a specific Pareto optimal solution, then you need to solve a single-objective problem obtained by scalarizing (aggregating) the objectives. The type of scalarization and values of its parameters (if any) depend on decision maker's preferences, e.g. how he/she/you want(s) to prioritize different objectives when they conflict with each other. Weighted sum, achievement scalarization (a.k.a. weighted Chebyshev), and lexicographic optimization are the most widespread types. They have different advantages and disadvantages, so there is no universal recommendation here.
CPLEX is preferred in the case, where (A) your scalarized problem belongs to the class solved by CPLEX (obviously), e.g. it is a [mixed integer] linear/quadratic problem, and (B) the problem is complex enough for computational time to be essential. CPLEX is specialized in the narrow class of problems, and should be much faster than Matlab in complex cases.
You do not have to limit the choice of multiobjective methods to the ones offered by Matlab/CPLEX or other solvers (which are usually narrow). It is easy to formulate a scalarized problem by yourself, and then run appropriate single-objective optimization (source: it is one of my main research fields, see e.g. implementation for the class of knapsack problems). The issue boils down to finding a suitable single-objective solver.
If you want to obtain some general information about the whole Pareto optimal set, I recommend to start with deriving the nadir and the ideal objective vectors.
If you want to derive a representation of the Pareto optimal set, besides the mentioned population based-heuristics such as GAs, there are exact methods developed for specific classes of problems. Examples: a library implemented in Julia, a recently published method.
All concepts mentioned here are described in the comprehensive book by Miettinen (1999).
Can cplex solve a pareto type multiobjective one? All i know is that it can solve a simple goal programming by defining the lexicographical objs, or it uses the weighted sum to change weights gradually with sensitivity information and "enumerate" the pareto front, which highly depends on the weights and looks very subjective.
You can refer here as how cplex solves the bi-objetive one, which seems not good.
For a true pareto way which includes the ranking, i only know some GA variants can do like NSGA-II.
A different approach would be to use a domain-specific modeling language for mathematical optimization like YALMIP (or JUMP.jl if you like to give Julia a try). There you can write your optimization problem with Matlab with some extra YALMIP functionalities and use CPLEX (or any other supported solver as a backend) without restricting to one solver.