Dears,
I realized that 99% of the time my problems of optimization have ONLY a huge number of boolean decision variables(400K) and linear constraints. I'm wondering if I can increase the performance of CP-SAT avoiding unnecessary solver actions using some of the parameters listed here:
https://github.com/google/or-tools/blob/master/ortools/sat/sat_parameters.proto
Thanks!
p.s. I'm using python
You can try:
linearization_level:0
or
num_search_workers:8
Related
I am new to LP and have only briefly used PuLP in Python.
Why is there a speed difference between SCIP 3.2.1 - CPLEX 12.63 and CPLEX 12.6.3? Doesn't SCIP still use CPLEX for solving?
Why will someone use SCIP with CPLEX solver, instead of using CPLEX directly?
What is this difference about
This plot is not showing a LP-benchmark, but a Mixed-integer programming benchmark.
Mixed-integer programming solvers typically use a branch-and-cut-based algorithm (including heuristics and co.), where a lot of relaxations are solved (in sequence; treating binary-/integer-variables as continuous resulting in an LP-problem).
One decision then is to choose how to solve these relaxed subproblems. The most simple decision (there are many more; e.g. tuning the Simplex-algorithm's parameters; it get's even more complex when solving problems with nonlinear-conic objectives) is to choose the LP-solver.
SoPlex is a LP-solver implementation by the SCIP-team. Meaning:
SCIP - SoPlex will use SCIP's algorithm for MIP (handling branching, cut-generation and co.) using SoPlex as solver for the internal LP-subproblems
SCIP - CPLEX will use SCIP's algorithm for MIP using CPLEX as solver for the internal LP-subproblems
Why using SCIP with CPLEX (instead of using a pure CPLEX approach)
The why is not that easy to explain.
Keep in mind, that all MIP-solvers are heuristics-based and on some problems SCIP will be faster than CPLEX (despite the underlying LP-solver selected).
Keywords for some theory: NP-hardness (of MIP) and the No free lunch theorem
Faster could mean: faster due to the MIP-based strategies, not the speed of the underlying LP-solver so that you may even gain an overall speedup using CPLEX on the subproblems!
The two solvers (MIP-solvers) are probably also much different in regards to parameters & accessibility (of internal algorithmic components). It's obvious, that you can tune SCIP in a much more general way than CPLEX (because it's open source)
As mattmilten mentioned in the comments: SCIP and CPLEX are also different in regards to the support of problem-classes which can be solved. One example of this might be the possibility for some special nonlinear-constraints (resulting in a MINLP). Using SCIP for these kind of problems, can still use CPLEX' LP-solver internally (same arguments as above)
I am using CPLEX for MATLAB toolbox wherein I formulate my MILP as a huge matrix and use the function cplexmilp to call the solver. Since the model I am solving is really huge, I intend to set the option of using multiple processors to speed up the solving of the MILP. I went through the manual but I could not find any specific solution for my case. I'd be grateful if anyone could help me find a solution to this issue.
CPLEX Toolbox Function
Thanks Everyone
I'm using MATLAB 2010b 64bit and its cplex integration to solve an engineering problem. However, because of the memory leak of cplex, memory usage exceeds acceptable limits with cplex (100+GBs including virtual memory) hence I am not able to solve my problem. You can see a similar post here.
Then I tried to use MATLAB linprog from the optimization toolbox but the result was disappointing. The running time of the algorithm for a small problem instance was increased from 80 cpu sec to 2600 cpu sec.
Now, I need an LP solver integration to MATLAB which is similar to CPLEX or linprog. By "similar" I mean the way it accepts data input in the form (F, A, B, Aeq, Beq, ...etc).
I must be able to use it in loops. Do you have any suggestions for that?
I would be very surprised if there was a memory leak in cplex. If you have a large problem then the memory will grow with any sensible solver. Is there perhaps a memory leak in the interface to cplex? How big is your problem? Are you running multi threaded as each thread will take a copy of the problem and hence will eat a lot more memory.
You should not be surprised to find that other solvers take a lot longer than cplex to solve your problem. Certainly the free solvers will be very much slower than cplex for any large problem.
After some trials to fix MATLAB/CPLex API's memory usage problem (memory leak) and after referring to some studies I decided to switch to Gurobi solver. For pure LP problems, it seems to be slightly slower compared to CPlex but this can be due to the way I use Gurobi. Someone may find Gurobi faster compared to CPlex. I suggested that on my previous posts under different questions. Here are some academic studies[Analysis of commercial and free and open source solvers for linear optimization problems][1]
[1] : http://www.statistik.tuwien.ac.at/forschung/CS/CS-2012-1complete.pdf
I am fairly new to Mixed Integer Linear Programming and I was hoping someone could clarify a performance question for me. Basically I am performing a calculation with about 34 decision variables and my calculation time is around 5 seconds. I would like to ideally get the calculation time down into the sub 1 second range.
Currently I am using the CBC solver & MATLAB, but as I understand it this is a single-threaded solver. Most of the MILP solvers I've seen seem to pride themselves on large project performance with 1k+ variables and knocking compute time down from days to hours but only a few of the expensive ones are even multi-threaded. Processor speed would seem to only go so far with such a problem so there has to be something that could be done on the software side.
In a situation like I have what factors play a role in the calculation time? In theory would a solution like Gurobi be able to ramp up and make a discernible difference over CBC on such a small problem?
When solving MIPs, multi-threading usually doesn't help that much, compared to other applications. One thing more sophisticated solvers do better than CBC is presolving. It may very well happen, that all decision variables can be eliminated/fixed in the root node. In general, it is not possible to predict the solving time of a MIP based on its size. The solving process is simply too complex.
To answer your question: I do suppose that you will see a significant speedup when trying another solver (Gurobi, Xpress, CPLEX, SCIP, etc.). It is not guaranteed though, and you might also witness a slowdown.
Most of the commercial MILP solvers would probably consider most problems with only 1000 variables to be tiny. It is very common to solve problems with hundreds of thousands or millions of variables. Note however, that there are also some really small problems that are considered to be very hard or still open in that nobody has solved them to a proven optimal solution. Basically, the time taken to solve these problems is enormously problem dependent.
If you want to see how fast another solver can solve your problem, try submitting it as e.g. an MPS file to the NEOS solver service at http://www.neos-server.org/neos/
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.