The remathematization of simulations (from 1998 onwards)

• Published in: From Models to Simulations pp. 118 - 142
• Main Author: Varenne, Franck
• Format: Book Chapter
• Language: English
• Published: United Kingdom Routledge 2019; Taylor & Francis Group
• Edition: 1
• Subjects: Statistical Model Schema; Architectural Simulation; AMAP; Plant Morphogenesis; Computer Infrastructure; Monte Carlo Type Simulation; Louis Pasteur University; Rewriting Rules; ENSAT; Silicon Graphics IRIX; Silicon Graphics Workstations; SSM; Cellular Automata; Elementary Time Steps; Turtle Geometry; Structural Functional Model; Multi Level Model; Stochastic Simulation; DLA; Virtual Laboratory; Common Language; Increased Computing Complexity; Hydric Potential
• ISBN: 9780367586621; 0367586622; 1138065218; 9781138065215
• DOI: 10.4324/9781315159904-8

The previous chapter suggested that an all-simulated solution was no longer tenable in cases where the aim was to integrate the plant's physiology into the computer simulations. The CIRAD laboratory, which in the period from 1998 to 2002 had been headed by François Houllier, had produced work tending to simplify the original simulations without relinquishing their underlying principles, but at the same time introducing new mathematical sub-models that - at certain stages and in a simplified manner - would deal with the physiological functioning of the plant. Starting from this initial simplification, de Reffye persevered, aided first by the expertise in automation of a laboratory in China, and later by French researchersfrom INRIA. The establishment of the joint Greenlab CIRAD/INRIA/École Centrale Paris1 team ultimately formalized the entry into this "remathematization" phase, as I call it, resulting essentially in a belated convergence between CIRAD's approach and the school of algorithmic simulation inspired by Lindenmayer. We will see, thus, that a relatively mature simulation calibrated on a large number of data and, in particular, the implementation of this simulation by computer, provided a formal middle ground on which it was possible to organize in return a simplification and a symbolic standardization - or, in other words, a remathematization. It is significant that, in this context, it was not a concern for conceptualization or comprehension imposed on simulation as a result of outside demand from theoretical biologists, for instance; instead, essentially for reasons of feasibility (regarding calculation times and memory capacities) and of verifying the simulations, it was the computer itself that ultimately needed to understand what it was doing, so to speak. The simulation remathematization work carried out by the Digiplante laboratory was based on the complexity of the AMAP simulations, but proposed a "structural factorization of the plant" that played a part in relinquishing the simulation solution. The simulation was still governed by the fragmented model of architectural growth at each stage. The model was purely computer-based. It considered the tree as "a collection of simple units corresponding to the organs". Fractal - essentially the mathematics of the concrete - were having a significant impact in the rapidly expanding computer graphics circles of that time, even though in the case of plant morphogenesis their direct use was limited to a few species of fern. Philippe de Reffye considered the Beijing partners' approach to be more practical and less biased on the subject of modelling and simulation. During the 1980s and 1990s, new, purely physicalist models also regularly appeared, in pace with advances in physical chemistry.