Thilo Moshagen

Thilo Moshagen

In engineering, it is a common desire to couple existing simulation tools together into one big system by passing information from subsystems as parameters into the subsystems under influence. As executed at fixed time points, this data exchange gives the global method a strong explicit component. Globally, such an explicit cosimulation schemes exchange time step can be seen as a step of an one-step method which is explicit in some solution components. Exploiting this structure, we give a convergence proof for such schemes. As flows of conserved quantities are passed across subsystem boundaries, it is not ensured that systemwide balances are fulfilled: the system is not solved as one single equation system. These balance errors can accumulate and make simulation results inaccurate. Use of higher-order extrapolation in exchanged data can reduce this problem but cannot solve it. The remaining balance error has been handled in past work by recontributing it to the input signal in next coupling time step, a technique labeled balance correction methods. Convergence for that method is proven. Further, a proof for the lack of stability of such methods is given for cosimulation schemes with and without balance correction.

Thilo Moshagen, Nihal Acharya Adde, Ajay Navilarekal Rajgopal

The logcosh loss function for neural networks has been developed to combine the advantage of the absolute error loss function of not overweighting outliers with the advantage of the mean square error of continuous derivative near the mean, which makes the last phase of learning easier. It is clear, and one experiences it soon, that in the case of clustered data, an artificial neural network with logcosh loss learns the bigger cluster rather than the mean of the two. Even more so, the ANN, when used for regression of a set-valued function, will learn a value close to one of the choices, in other words, one branch of the set-valued function, while a mean-square-error NN will learn the value in between. This work suggests a method that uses artificial neural networks with logcosh loss to find the branches of set-valued mappings in parameter-outcome sample sets and classifies the samples according to those branches.

Nihal Acharya Adde, Thilo Moshagen

In the case of clustered data, an artificial neural network with logcosh loss function learns the bigger cluster rather than the mean of the two. Even more so, the ANN when used for regression of a set-valued function, will learn a value close to one of the choices, in other words, it learns one branch of the set-valued function with high accuracy. This work suggests a method that uses artificial neural networks with logcosh loss to find the branches of set-valued mappings in parameter-outcome sample sets and classifies the samples according to those branches. The method not only classifies the data based on these branches but also provides an accurate prediction for the majority cluster. The method successfully classifies the data based on an invisible feature. A neural network was successfully established to predict the total number of cases, the logarithmic total number of cases, deaths, active cases and other relevant data of the coronavirus for each German district from a number of input variables. As it has been speculated that the Tuberculosis vaccine provides protection against the virus and since East Germany was vaccinated before reunification, an attempt was made to classify the Eastern and Western German districts by considering the vaccine information as an invisible feature.

Thilo Moshagen

In engineering, it is a common desire to couple existing simulation tools together into one big system by passing information from subsystems as parameters into the subsystems under influence. As executed at fixed time points, this data exchange gives the global method a strong explicit component, and as flows of conserved quantities are passed across subsystem boundaries, it is not ensured that systemwide balances are fulfilled: the system is not solved as one single equation system. These balance errors can accumulate and make simulation results inaccurate. Use of higher-order extrapolation in exchanged data can reduce this problem but cannot solve it. The remaining balance error has been handled in past work with balance correction methods which compensate these errors by adding corrections for the balances to the signal in next coupling time step. Further past work combined smooth extrapolation of exchanged data and balance correction. This gives rise to the problem that establishing balance of one quantity a posteriori due to the time delay in general cannot establish or even disturbs the balances of quantities that depend on the exchanged quantities, usually energy. In this work, a method is suggested which allows to choose the quantity that should be balanced to be that energy, and to accurately balance it.