Computing with Real Numbers

This is a continuation of the previous post How to avoid collapse of modern mathematics.

Let me see if the constructive computational approach to mathematics adopted in the BodySoul program can meet the criticism expressed by Norman Wildberger as concerns the foundations of the large areas of mathematics relying on the concept of real number.  In particular Wildberger asks about the elementary process of adding two real numbers such as $\sqrt{2}$ and $\pi$: 

• $\sqrt{2}+ \pi = ?$ 

Let us then use the least cryptic definition of a real number as an infinite decimal expansion. But asking for the infinite decimal expansion of $\sqrt{2}$ is asking too much, and so we have to limit the specification to a finite number of decimals, and the same with $\pi$. We can then add these numbers using well specified rules for computing with rational numbers, and so arrive at a finite decimal expansion as an approximation of $\sqrt{2}+ \pi$. We can choose the number of decimals to meet a given precision requirement. Fair enough. 

But how do we know the decimal expansions of $\sqrt{2}$ and $\pi$?. Before the computer they would have to be picked up from a printed precomputed mathematical table, but only up to finitely many decimals and the table would swell beyond all limits by asking for more and more decimals. Today with the computer, you can press a button and let $\sqrt{2}$ be computed from scratch using Newton's method, but even if this algorithm is very efficient, the required work/time would increase beyond limit by asking for more and more decimals. 

The computer would compute the sum $\sqrt{2}+ \pi$ in an iterative computational process involving:

1. Compute $\sqrt{2}$ with say $5$ decimals.
2. Compute $\pi$ with say $5$ decimals.
3. Add these decimal expansions using the addition algorithm for finite decimal expansions.
4. Check if a desired precision is met, and if not go back to 1. and increase the number of decimals.  

This would reduce the foundation of mathematics to computational processes, and this is the approach of BodySoul: All mathematical objects are constructed by specified finitary computational processes as finite precision solutions to specified equations. 

For example, the value of the exponential function $\exp(t)$ for any value $t>0$ is computed by solving the differential equation $x^\prime (s)=x(s)$ for $0<s\le t$ with $x(0)=1$ by time stepping, where $x^\prime $ is the derivative of $x$, and setting $\exp(t)=x(t)$. No values of $\exp(t)$ are stored. New computation from scratch for each value of $t$. This is the only way to avoid storing real numbers as infinite decimal expansions, which is impossible in a finite Universe. 

Is Wildberger happy with such a response to his criticism. And what about you?

In any case, pure mathematicians will not welcome a foundation based on non-pure computational mathematics, even if it would solve unresolved foundational questions concerning real numbers and elementary functions of real numbers as solutions to differential equations. 

There was a tough fight at the turn to to modernity in the beginning of the 20th century concerning the foundations of mathematics between logicism (Russell), formalism (Hilbert) and constructivism/intuitionism (Brouwer), which was won by Hilbert in the 1930s thus setting the scene for 21st century mathematics. But with the computer, constructivism is now taking over by offering a concrete foundation without lofty speculation of infinities.  

A formalist can introduce $\mathcal{R}$ as the set of equivalence classes of all Cauchy sequences of rational numbers thus as a set defined by a certain property. Russell showed the danger of defining sets this lofty way by his famous example of a set defined by the property of not containing itself leading to a contradiction. Gödel turned Russell's example into more precise form, which should have killed both the logicist and formalist school, but did not since the reaction was to kick out constructionists compatible with Gödel from mathematics departments to form separate departments of computer science. Mathematics departments/education is still controlled by formalists, which means that Wildberger's criticism is not welcome.