What is Wolfram Language (mathematica)'s influence on Julia?

5 posts were split to a new topic: On flagging newcomers as off-topic

One of my favorite quotes of all time is from Ed Spiegel, a pioneer in turbulence and chaos:

If I have seen less far than others, it is because I was standing behind giants.


Haha I love this quote! I cannot find a real reference for it, so please do share if you have. I think it is really profound, since sometimes being too focused on what others has done in the past, hinders your own ideas :slight_smile:

I would suggest mathmatica has very little influence these days because it is not very transparent. It’s all closed source. People in for example, well known corporation Microsoft, created, say, for example D*, which does mathematica faster than mathematica does.

So a better question it ask is what do you care about? Generalization? Speed? Scale? Money?

At least in the modern theory of computer science people have learned to embrace the kripean semantics that naming things = creation of bias, which in that school of thought that you can just autocompute all the things, combinatorial branch and bound optimization is still very hard, but it depends on where you apply it (for example in microgrid optimization of power). Drop the necessary condition, and learn how to fix defects automatically. Save energy and see information theoritic “bits” of entropy as pollution you can “fix” if you apply reflexivity properly.

Sorry for the abstract language, just create positive feedback loops in your own lines of your own tunnel visioned and you can see what I and other people had apparently been thinking.

The reference is http://chaosbook.org/chapters/intro.pdf, page 2. The quote is attributed to “Eduardo Specchio”, which is the author Predrag Cvitanović’s tongue-in-cheek name for Ed Spiegel.


Cool book! To me, to understand the initial conditions or boundary conditions of whatever fractal dimensions can be thought differently. You can also frame questions using say breaking of symmetries and do lie integration in coexter groups and look and just count stuff. One good meta algorithm is when one problem sounds hard do the other problem, but there may be many stable manifold ways to see equivalence patterns between P=NP, Continuum dynamics and all of the dualities expressed in modern mathmatical physics. Luckily these days you can virtualize creating your dual diamond cutters and confine a particle beam to perform some controlled quasi monte carlo algo to compare some log odds and examine the rich structure of say the p-adic numbers using 1 second of wall clock time. Instead of say looking at the statistical mechanical properties of some continuum mechanics despite all the gauge theories that need local/global laws to do crypto quantum information theoretic tests that must exist because we still can’t integrate matter despite deformation theory and renormalization groups. I think categorification (creating endofunctors, and assuming everything is defined by it’s effects on everything else) is very useful since it allows thinking in very general systematic ways. Then you can start numerically computing in modern pure math terms and use algebraic geometry to deloop stuff.

Goodness, are you trying to break some sort of record on the number of multisyllabic words you can cram into a post? You might find better success at engaging with folks if you can make things a little less abstract.


I think Scheme literally influenced Julia more than CLOS, but apparently we get biased by exposure to our favorite tools. Remember, historical parsimony happens. For example, C might have only been popular because AT&T gave away shit for free. Bell Labs was a monopoly till AT&T was broken up. But all the communications networks it built allowed people to collaborate. C++ which subsumed C in the early 90s because say SGI proposed the STL which sounded like a good idea. Meanwhile, Sun was working on academic languages and eventually came out with the JVM. There was basically an anti-JVM created by a spawn of apple called (TypeScript) which was much more modular (basically it was a anti-JVM) and made to be power saving and allowed distributed compute. But it was never created, but the company (General Magic) was very influential in the Silicon Valley. Many ideas happen to be created by accident by connecting the dots. This is referred to the generalized Ramsey theory bias. David Hilbert and his “school” of mathmatics created inf dim linear spaces. But maybe that’s not the best way of thinking about it. It’s just one common way of coming to the same conclusion.

One meta algorithm is that generally most algos just are doing some splitting or utilizing some sparacity pattern. If you study VC-dimensions (statistical machine learning), you’ll understand why. Better IMHO to understand renormalization and the general theory of relativity and the calculus of arbitrary differences across arbitary finite ends and coends. It makes easy to compute all the stuff with arbitrary physics and do einstein-like thought experiments on a computer in safe spaces. Create some theories of special relativity and change Relus to dirac 'probe" functions. Then figure out what localized field effect this “probe” can do a gauge theory. Remember all the fields of human experience are just abstractions on whatever fields. For example, Danzig was very influenced by just meeting Von Neuman or Feynman. We are in a different era of computer architecture with software defined hardware. You can save labor by computing by creating bridge adapters between stuff that is already computed. Then you can create the tight topological spaces where good adjoints can be created. It’s just a metaphor that cognitive scientists think these days relates to how our own brain works. Apparently we can telepath info to each other digitally via the parietal lobe. So it’s best to learn how to thicken yourself and create value by identifying antipatterns (which you can learn about) and best practices.

For example, think about the difference between irradiance and filament bulbs and dark and antidark. What saves more energy? SpaceX will soon create huge more satellite bandwidth by also having the satellites feel each other. All your devices can triangulate each other locally. These are powerful energy saving tricks but prepare for a much faster moving technologically band gap international conversion of coding apps.

This is not abstraction, it is noise. (Here and in other topics as well, unfortunately.)


I am sorry but I could not make sense of your post, or even decide if you are arguing the above.

In any case, since Julia is turning out to be successful, I imagine that a lot of people will claim that various ideas and languages influenced it, back to and including Babylonian clay tablets. This is normal.

I think it is best to just ask the originators of Julia about this. Or be conservative and just assume that the languages in the Why we created Julia announcement are the relevant ones.

Taking the “X influenced Y” game to its extremes can become meaningless pretty quickly. My favorite example is from pop music: