Our Current Projects

RefFit

RefFIT has a powerful and fast fitting engine based on a modified Levenberg-Marquardt algorithm, a broad collection of proven physical models and a unique user-friendly interface. It allows you fitting manually or automatically several datasets of different type and using different models simultaneously while seeing on your screen precisely what you are doing.

Now we plan to develop a new functionality, which is to write a routine which can automatically generate analytical derivatives of all mathematical functions such that each scientific model from the library can be combined mathematically with others while the program automatically generates and tracks the analytical derivatives of the models. In this way, one can, with the use of a limited set of preprogrammed models quickly created any type of complex, nonlinear model. 

With this analytical code, complex data analysis is faster and more reliable than using numerical solutions, like using standard numerical fitting libraries in R or Python. In addition, due to the automatically generated models, coding by the users will be redundant. Therefore, the expensive and lengthy process of development of models either by scientists, a consultant or in-house by a company can be eliminated. We can thus propose a data science software which delivers fast and reliable fitting to large datasets tailor-made for the user without any need for coding. 

 

https://reffit.ch

 

 

The team

Iris Crassee

Senior research and teaching assistant

 
 
 
 
 
January 8, 2021
  Our Current Projects