I need to repeatedly compute derivatives at a given point of nonlinear pyomo constraints. According to this post, there are basically two options: one symbolic approach (which uses sympy) and one numeric approach via the NL file generated by Pyomo.
I have tried the symbolic approach which looks like:
def gradient(constr, variables):
grad_num = np.array([value(partial) for partial in
differentiate(constr.body, wrt_list=variables)])
if (not (is_leq_constr(constr))):
grad_num = -grad_num
return grad_num
In principle, this approach works. However, when dealing with larger problems, the function call takes very long and I would expect the numeric approach to be much faster.
Is this true? If so, can anyone help me, to "translate" the above function to the gjh_asl_json approach? I am grateful for any hint.
Related
Suppose we're given some sort of graph where the feasible region of our optimization problem is given. For example: here is an image
How would I go on about constructing these constraints in an integer optimization problem? Anyone got any tips? Thanks!
Mate, I agree with the others that you should be a little more specific than that paint-ish picture ;). In particular you are neither specifying any objective/objective direction nor are you giving any context, what about this graph should be integer-variable related, except for the existence of disjunctive feasible sets, which may be modeled by MIP-techniques. It seems like your problem is formalization of what you conceptualized. However, in case you are just being lazy and are just interested in modelling disjunctive regions, you should be looking into disjunctive programming techniques, such as "big-M" (Note: big-M reformulations can be problematic). You should be aiming at some convex-hull reformulation if you can attain one (fairly easily).
Back to your picture, it is quite clear that you have a problem in two real dimensions (let's say in R^2), where the constraints bounding the feasible set are linear (the lines making up the feasible polygons).
So you know that you have two dimensions and need two real continuous variables, say x[1] and x[2], to formulate each of your linear constraints (a[i,1]*x[1]+a[i,2]<=rhs[i] for some index i corresponding to the number of lines in your graph). Additionally your variables seem to be constrained to the first orthant so x[1]>=0 and x[2]>=0 should hold. Now, to add disjunctions you want some constraints that only hold when a certain condition is true. Therefore, you can add two binary decision variables, say y[1],y[2] and an additional constraint y[1]+y[2]=1, to tell that only one set of constraints can be active at the same time. You should be able to implement this with the help of big-M by reformulating the constraints as follows:
If you bound things from above with your line:
a[i,1]*x[1]+a[i,2]-rhs[i]<=M*(1-y[1]) if i corresponds to the one polygon,
a[i,1]*x[1]+a[i,2]-rhs[i]<=M*(1-y[2]) if i corresponds to the other polygon,
and if your line bounds things from below:
-M*(1-y[1])<=-a[i,1]*x[1]-a[i,2]+rhs[i] if i corresponds to the one polygon,
-M*(1-y[1])<=-a[i,1]*x[1]-a[i,2]+rhs[i] if i corresponds to the other polygon.
It is important that M is sufficiently large, but not too large to cause numerical issues.
That being said, I am by no means an expert on these disjunctive programming techniques, so feel free to chime in, add corrections or make things clearer.
Also, a more elaborate question typically yields more elaborate and satisfying answers ;) If you had gone to the effort of making up a true small example problem you likely would have gotten a full formulation of your problem or even an executable piece of code in no time.
I have an optimization problem where I want to minimize for the total cost of a system, so I write an objective function that is the sum of my different costs. The problem includes using one of three machines each one with different cost at a different threshold of usage. I define each machine (model.Machine#) as a binary variable and declare the parameters of each machine cost model.Cost#). I am trying to get the cost to be able to minimize it but when I write the constraint:
model.Cost1*model.Machine1 + model.Cost2*model.Machine2 + model.Cost3*model.Machine3 == model.MachineCost
Where I also write:
model.Machine1 + model.Machine2 + model.Machine3 == 1
Gurobi is telling me that it can't handle an quadratic function referring to the first constraint mentioned above. However it is parameters multiplied by binary variables there isn't anything quadratic.
I know the question is vague and part of a larger problem but I hope you can understand what I am referring to and help me!
Thank you so much for your assistance!
What is model.MachineCost? Is it an Expression component with some kind of quadratic expression stored inside of it?
If not, can you start commenting out things in your model until you get down to a minimal working example (that causes this error) and post that? Otherwise, we can't be sure that there are not other quadratic pieces of the model that you are not showing.
Is there a function in Julia that is similar to the solver function in Excel where I can provide and equation, and it will solve for the unknown variable? If not, does anybody know the math behind Excel's solver function?
I am not expecting anybody to solve the equation, but if it helps:
Price = (Earnings_1/(1+r)^1)+(Earnings_2/(1+r)^2)++(Earnings_3/(1+r)^3)+(Earnings_4/(1+r)^4)+(Earnings_5/(1+r)^5)+(((Earnings_5)(RiskFreeRate))/((1+r)^5)(1-RiskFreeRate))
The known variables are: Price, All Earnings, and RiskFreeRate. I am just trying to figure out how to solve for r.
Write this instead as an expression f(r) = 0 by subtracting Price over to the other side. Now it's a rootfinding problem. If you only have one variable you're solving for (looks to be the case), then Roots.jl is a good choice.
fzero(f, a::Real, b::Real)
will search for a solution between a and b for example, and the docs have more choices for algorithms when you don't know a range to start with and only give an initial condition for example.
In addition, KINSOL in Sundials.jl is good when you know you're starting close to a multidimensional root. For multidimensional and needing some robustness to the initial condition, I'd recommend using NLsolve.jl.
There's nothing out of the box no. Root finding is a science in itself.
Luckily for you, your function has an analytic first derivative with respect to r. That means that you can use Newton Raphson, which will be extremely stable for your function.
I'm sure you're aware your function behaves badly around r = -1.
I am solving an optimization problem. the problem has binary constraints. solver is (during iteration) setting those binary constraints to decimals between 0 and 1 (approximating a relaxed gradient search). I wish to indicate to solver that it should just search over the discontinous values for 0..1.
Is there a way to do this?
Alternatively, is there an algorithm in OpenSolver which does this, that mimics the simplex-lp, and provides a global optimum?
the cheap way to do it, is to right a for-loop, and iterate over the values. I was wondering if there was a way to phrase it so that a nonlinear problem, becomes a linear problem.
Thanks.
The GRG Nonlinear and Simplex LP methods both use the Branch & Bound method when faced with integer constraints. This method "relaxes" the integer requirement first, finds a solution, then fixes one of the constraints to an integer and finds a new solution. See the Solver on-line documentation.
It is a brute force search method and can take a considerable amount of time.
The Evolutionary method uses it's own algorithm for dealing with integer constraints and is typically much faster than the other two methods.
You ask about linearizing a non-linear problem - you would need to provide more specific information in order to answer that (e.g. What is your equation? How have you set up your solver problem? etc.)
CLP(FD) allows user to set the domain for every wannabe-integer variable, so it's able to solve equations.
So far so good.
However you can't do the same in CLP(R) or similar languages (where you can do only simple inferences). And it's not hard to understand why: the fractional part of a number may have an almost infinite region, putted down by an implementation limit. This mean the search space will be too large to make any practical use for a solver which deals with floats like with integers. So it's the user task to write generator in CLP(R) and set constraint guards where needed to get variables instantiated with numbers (if simple inference is not possible).
So my question here: is there any CLP(FD)-like language over reals? I think it could be implemented by use of number rounding, searching and following incremental approximation.
There are at least some major CLP(FD) solvers that support real (decision) variables:
Gecode
JaCoP
ECLiPSe CLP (ic library)
Choco (using Ibex)
(The first three also support var float in MiniZinc.)
The answser to your question is yes. There is Constraint-based Solvers dedicated for floating numbers. I do not have a list of solvers but I know that that ibex http://www.ibex-lib.org is a library allowing the use of floats. You should also have a look at SMT-Solvers implementing the Real-Theory (http://smtlib.cs.uiowa.edu/solvers.shtml).