Replicating and breaking models: good for you and good for ecology
Replicating and breaking models: good for you and good for ecology
复制标题
DOI：
10.1111/oik.02170
复制
发表时间：
2015-06
期刊：
影响因子：
3.4
通讯作者：
Jan C. Thiele;V. Grimm
中科院分区：
文献类型：
--
作者：
Jan C. Thiele;V. Grimm
生态学中的计算模型产生一般性见解的潜力有两个主要限制：它们的设计是路径依赖的，反映了不同的潜在问题，假设和数据，并且很少有鲁棒性分析探索解释某些观察结果的模型机制。我们在这里认为，这两个限制可以克服，如果生态学建模者将更经常地复制现有的模型，试图打破模型，并探索修改。复制包括现有模型的重新实现及其结果的复制。打破模型意味着确定在什么条件下模型中表示的机制不再能解释观察到的现象。复制的好处包括进入模型开发的迭代阶段所花费的精力更少，并且有更多的时间进行系统的健壮性分析。复制文化将导致提高可信度，一致性和计算模型的效率，从而促进理论的发展。
There are two major limitations to the potential of computational models in ecology for producing general insights: their design is path-dependent, reflecting different underlying questions, assumptions, and data, and there is too little robustness analysis exploring where the model mechanisms explaining certain observations break down. We here argue that both limitations could be overcome if modellers in ecology would more often replicate existing models, try to break the models, and explore modifications. Replication comprises the re-implementation of an existing model and the replication of its results. Breaking models means to identify under what conditions the mechanisms represented in a model can no longer explain observed phenomena. The benefits of replication include less effort being spent to enter the iterative stage of model development and having more time for systematic robustness analysis. A culture of replication would lead to increased credibility, coherence and efficiency of computational modelling and thereby facilitate theory development.