Mancosu - Philosophy of Mathematical Practice (Oxford, 2008).pdf

mathematical concepts: fruitfulness and naturalness 293applications in physical science and common-sense reasoning about the physicalworld. (Of course, in some cases there will be close connections, but this isn’tnecessary to the point.)¹² Whether or not the conjectures in question haveany obvious connection to any direct applications makes no difference to ourassessment of the patterns of reasoning informing the making and supportingof conjectures. It may well be that in an ideal epistemology, the status ofmathematics as knowledge must ultimately appeal to physical applications, or itmay not; we can be neutral on that point here.10.4 Fruitfulness, conjecture, and RiemannianmathematicsThe Riemann–Dedekind methodology mentioned in the introduction incorporatesexactly the connection between identification of core propertiesand mathematical fecundity discussed in the previous section. In this section I’llsay a bit more about the Riemann–Dedekind approach. First consider theseRiemannian remarks. They seem innocuous and even dull, but understood incontext they are the first shot in a revolution.Previous methods of treating [complex] functions always based the definition ofthe function on an expression that yields its value for each value of the argument.Our study shows that, because of the general nature of a function of a complexvariable, a part of the determination through a definition of this kind yields therest ... This essentially simplifies the discussion ...A theory of these functions on the basis provided here would determine thepresentation of a function (i.e. its value for every argument) independently of itsmode of determination by operations on magnitudes, because one would add tothe general concept of a function of a variable complex quantity just the attributesnecessary for the determination of the function, and only then would one goover to the different expressions the function is fit for. The common characterof a class of functions formed in a similar way by operations on quantities, isthen represented in the form of boundary conditions and discontinuity conditionsimposed on them. (Riemann, 1851, p.38)Riemann’s point is simple. We are wondering how to characterize a wellbehavedcomplex function or a class of them. How should we proceed? One¹² Though I’m not exploring the point here, a closeness to physical applications was in fact recognizedas a feature of Riemann’s mathematics. ( This is touched on in the opening quotation from Stahl.)For example, Helmholtz found Riemann’s complex analysis congenial, presumably in part becauseRiemann’s classification of functions by their singularities fit smoothly with Helmholtz’s approach topotential flow (cf. Darrigol, 2005, p.164).