Computer science data import tools

Whereas Bernal had to rely primarily on mechanically produced random sphere packings in his work on liquid structure, the molecular dynamics and Monte Carlo computer simulation techniques have, during the past three decades, provided researchers with new and powerful experimental tools enabling a much closer look into the structure of the liquid—one has available the trajectory of every atom. Despite this, computer simulation has been used principally to calculate liquid-state correlation functions. This situation, to quote Lumsden and Wilson [6], ... appears to stem in part from a peculiar and fundamental relation that has always existed between experiment and theory in science the importance of experimental data is judged by the theory to which it is applied. As the physicist Arthur Eddington said half seriously, no fact should be accepted as true until it has been confirmed by theory. Unless an attractive theory exists that decrees certain kinds of information to be important, few scientists will set out to acquire the information. Thus, it is only infrequently that computer simulations have been used to characterize liquid structure in ways other than those dictated by the prevailing liquid theory. 

This chapter is not the best place to illustrate the computer science aspects of the project. I will only mention that the Amsterdam team successfully developed a very sophisticated import tool, through which the individual contributors were able to check their spreadsheets for internal consistency and completeness quite easily by themselves, and eventually to correct them, before submitting the data to the technical staff of Fauna Europaea. 

A decade ago, artificial intelligence (AI) was mainly of interest to computer scientists. Few researchers in the physical sciences were familiar with the area fewer still had tried to put its methods to practical use. However, in the past few years, AI has moved into the mainstream as a routine method for assessing data in the experimental sciences it promises to become one of the most important scientific tools for data analysis. 

---