The ABC of optimization
API stands for Application Programming Interface. APIs are used to connect third-party programs to a software at source code level. During programing, APIs unify the data transfer between an application and individual program parts, i.e. the modules. The aim of APIs is to simplify programing through modularization. Individual modules are encapsulated from the actual code and communication is exclusively via the API.
Artificial Intelligence (AI)
Artificial Intelligence (AI) enables machines, in particular computing systems, to perform tasks autonomously, learn from experience and adapt their reactions accordingly – similar to human intelligence. Based on algorithms, explicit programming is not required for this adaptability. Machine Learning is a subfield of Artificial Intelligence.
Black box tuner vs. gray box tuner
A typical black box tuner only considers the end result of a target algorithm run – after solving the problem instance or when it has reached the time limit. The innovative gray box tuner applies intermediary results to predict the runtime. In this way, it aims to reduce the total runtime of the tuning and while still providing the same configuration quality as its black box counterpart. The OPTANO Algorithm Tuner supports both tuning approaches.
C# (c sharp) is a general object-oriented programming language which was developed as a native programming language for Microsoft’s .NET platforms in the early 2000s.
Descriptive Analytics is the examination of past data or content and serves to answer the question “What happened?” Available data from various systems and databases is prepared so that complexity is reduced. The result is reports, diagrams, dashboards, but also KPIs which help humans to interpret the information and support them in their decision-making processes.
Typical cases in which Descrpitive Analytics is applied are:
- KPI reporting
- Status monitoring
Diagnostic Analytics also works on the basis of historic data and seeks to answer the question: “What happened?” To do this it searches for causes and their effects on business processes in order to provide a detailed insight into specific problems. The data analysis helps to establish correlations and identify patterns. Domain knowledge is often required to draw conclusions from this and identify the correlations. Specialist knowledge such as entrepreneurship, industry know-how and methodological skills play a crucial role. Thus, deriving recommendations from the data is in the hands of experts.
Typical cases in which diagnostic analytics is applied are: :
- Identifying the effects of operational disruptions
- Detecting the causes of quality deficiencies
A digital twin is the digital representation of a real-world material or immaterial object or system. With the digital twin, processes can be analyzed, predicted and optimized. The aim is to achieve better business results and performance for a company and identify any problems early on. Digital twins are created from a data model, algorithms and knowledge.
The Gurobi Optimizer is a highly sophisticated solver for mathmetical programming. It solves optimization problems by implementing the latest algorithms. OPTANO has cooperated very well with Gurobi for many years now and is also a Gurobi Premium Partner.
The knapsack problem is a typical combinatorial optimization problem. The optimal combination of objects is to be assembled from a set, each with a defined weight and value. The goal is to maximize the value of the selcted objects. At the same time, this subset must not exceed the given weight limit.
Figuratively speaking, a person owns a knapsack with a maximum load capacity (e.g. in kilograms) as well as a number of objects, each with a weigt (also in kilograms) and a value (e.g. euros). By taking the load capacity of the knapsack into account, the value of the selected objects can be maximized.
Linear optimization / Linear programming
Linear optimization or linear programming is the most important sub-discipline of Operations Research and means the optimization of linear objective functions over a set which is constrained by linear equations and inequalties. The linear optimization models consist of linear objective functions which are minimized or maximized as well as number of linear constraints.
Machine learning is a subfield of Artificial Intelligence. Machine Learning enables IT systems to create knowledge in an automated manner and improve itself without being explicitly programmed to do so. Since Machine Learning recognizes patterns and correlations based on available datasets and the help of algorithms – which can also be trained – artificial knowledge is generated from experience so to speak. The insights gained in this way can be used to analyze unknown data, in order to make forecasts, solve problems and optimize processes.
Optimization in the mathematical sense means finding the optimal parameters of a – mostly complex – system. Mathematical optimization captures a business problem in a mathematical model and uses this to come up with the best solution.
.NET is a free-of-charge cross-platform open-source developer platform on which application programs can be created and performed.
Operations Research (OR)
Operations Research (OR) is an analytical method applied to solve problems and find optimal decisions. Using mathematical methods, OR supports decision-making processes, provides solutions to existing problems and prevents future problems from occurring.
OPTANO Algorithm Tuner
The OPTANO Algorithm Tuner is a .NET-API which supports the tuning of any target algorithm. In other words, near-optimal parameters are sought and found for this purpose. It does not matter here whether it is an OR-solver (e.g. Gurobi), a machine learning procedure or another algorithm. The OPTANO Algorithm Tuner is a flexible tool in many respects. It supports several award-winning tuning algorithms such as GGA, GGA++, JADE and CMA-ES and even the underlying tuning approach can easily be switched between the classic black box and gray box procedures.
OPTANO Modeling is the best API for mathematical programming in .net. It has been created by software developers who have already created hundreds of optimization applications. Since OPTANO Modeling meets software developers’ exact requirements, our module is therefore different to modeling systems which have been designed for mathematicians.
OPTANO Modeling is available as a NuGet Gallery package and can be downloaded free of charge.
The OPTANO Platform is a technology platform used to create optimization systems in planning. Whether it is the optimization of your supply chain, production or network – OPTANO makes planning processes efficient, transparent and flexible and supports you in your decision-making and planning processes using Prescriptive Analytics and what-if scenarios to provide sound recommendations and what-if scenarios.
OPTANO Predictive Blueprint
The OPTANO Predictive Blueprint is a Machine Learning feature which enables machine learning models to be created for various scenarios. These models can be used to predict future developments based on historic data.
The Predictive Blueprint is supported by ML.NET, an open-source platform for Machine Learning. A highly sophisticated feature has also been integrated using Auto ML. This means that high-performance algorithms are available for any forecast. Thus data can then be used as input for prescriptive models in integrated solutions.
OPTANO Production is an application specifically for production planning purposes. Whether tactical or operational planning – with OPTANO production you can create and adapt optimal plans quickly and easily.
OPTANO Web Client
The OPTANO Web Client was first launched with OPTANO 6.0. This web application has enabled us to extend the use of our optimization software even further since users can access their optimization projcts fast and easily – anytime and from anywhere.
Optimization solvers incorporate highly efficient algorithms to solve optimization models, therefore helping to improve decision-making in planning processes.
The objective of Predictive Analytics is to generate models from historical data and use these to predict unkown events or variations in the future. By forecasting future trends and states, it is possible to intervene at an early stage and optimize business processes. Predictive Analytics also requires domain knowledge to derive recommendations for action. However, predictions of processes and forecasted values create a transparency which makes it easier to weigh up the risks involved and creates an important foundation for decision-making.
Typical cases in which Predictive Analytics is applied are:
- Resource analysis
- Predictive maintenance
Prescriptive Analytics has the job of specifying decision alternatives in order to achieve projected results and, where necessary, to minimize future risks and maximize opportunity. It asks the question: “Which measures should be taken?” The objective of Prescriptive Analytics is to come up with with targeted recommendations on how to proceed or – as the highest degree of automation – to directly implement actions and optimizations in fully automated companies. Prescriptive Analytics thus indicates courses of action and forecasts the effects that specific decisions can have on the future, even though these decisions have not actually been selected. For this purpose it uses simulations, mathematical optimization, algorithms as well as machine learning techniques.
Typical cases in which Prescriptive Analytics is applied are:
- The procurement of goods to avoid supply bottleenecks and high inventory.
- Route planning for sustainable logistics.
Scenarios are possible images of a company in the future. They are based on the hypothetical sequence of events and their consequences with reference to a specific problem.
Scenarios contain all the relevant data required for analysis and optimization. Just like playing in a sandbox, data can be changed and different alternatives can be tested with no influence on the valuable productive data whatsoever.
see What-if Analysis
Traveling Salesman Problem (TSP)
The Traveling Salesman Problem (TSP) is an optimization problem in Operations Research and in theoretical Information Technology – a route planning problem, to be more precise. The task here is to find the shortest possible route so that a traveling salesman visits each city on his itinary exactly once and returns to the origin city at the end.
UX-Design stands for „User Experience Design“, wheras UI Design stands for „User Interface Design“.
UX-Design focus on the user’s experience as a whole and therefore also on usability. The aim of UX-design is to create efficient, intuitive and pleasant user experiences. UX-design looks at determining the relationships between elements as well as ensuring that navigation has been organized coherently.
UI-Design is a subfield within User Experience and contributes towards the usability of an application by graphically designing a user interface which enables the user to interact smoothly and intuitively. The layout of the user interface is the focus of UI-Design. It comprises all visual, interactive elements such as typography, icons, buttons, color schemes and responsive design.
In What-If Analyses or scenario analyses various future scenarios can be created and their effects can be considered in order to provide recommendations on the best action to take and find the best decisions.
The what-if analysis is a very good method to compare different solutions with one another and assess the risks involved in each. When comparing different solution approaches, scenarios and KPIs can be determined which support decision-making processes. With risk analysis, it is easier to estimate the potential risks and their impacts.