Sunday, June 30, 2013

My 1985 Notes on Wolfram's Harvard Talk on SMP - Precursor to Mathematica Part 2

Continuing from pp 1 - 3. This part contains some juicy tidbits that show how the principles underlying Wolfram's vision of a mathematical computer language have guided Mathematica's development to this day.

  

SMP contains the core of mathematical knowledge. A library of external files extends knowledge and gives definitions for particular applications; currently about 400 files.

Has a simple algorithm to strip to word roots and search a keyword index.

"This is a language, not something that thinks, and it was designed to be easy to tell things to it."

Dimensional analysis for physical units; fundamental units and derived terms' units.

Euler circuit in a graph (?)

Laplacian and divergences computed in various types of coordinate systems.

Routine mathematics should become as computerized as arithmetic.

SMP will run on a new generation of PCs.

One can typically define a hierarchy of types of objects. Then coerce two types to a common ancestor on a graph of types. How to get to the common ancestor? There may not be a unique path, so need a weighting function to [add efficacy to the search].

Vectorize: adds arrays of numbers in same number of operations as adding single numbers.

When to swap data with disc is left to the operating system. SMP needs a C compiler to work.

Gamma matrix manipulation.

No comments:

Post a Comment