HNHacker News
TopNewBestAskShowJobs

KenoFischer

4,368 karma · joined December 2, 2012

Julia Core Developer Co-Founder at Julia Computing

Reach me at keno@juliacomputing.com

submissionscomments
KenoFischer··on LLVM based Just-in-Time-Compiler for C++
I added a screenshot to the README.
KenoFischer··on LLVM based Just-in-Time-Compiler for C++
It does actually: http://imgur.com/eN9asVz. At some point I need to get around to creating a snazzy landing page for people not coming from the julia world, but who has the time ;).
KenoFischer··on LLVM based Just-in-Time-Compiler for C++
Cool! Is the code available anywhere? There's a couple of projects doing this and it would be good to share code. I assume this is using Clang? Do you have a list of things you'd like to see in upstream Clang? I'd imagine such a list of things would overlap quite a bit with what I or the cling folks need. If there's enough of a support base we might just be able to have upstream implement it.
KenoFischer··on On libunwind and dynamically generated code on x86-64
Oh hey, somebody else feels the pain! It's not only the crazy number of interfaces you have to support (not to mention that some of the interfaces are not particularly well defined, e.g. GDB's interface), but also what happens after. Most of these interfaces were not designed with very many JIT functions in mind and have O(n^2) behavior. One of the Julia core contributors recently worked with the GDB folks to fix their O(n^2), but libunwind still has it (to the point where one of the julia tests takes 15% of it's total runtime just having the unwinder try to find the JIT frame). Library support for JIT code is surprisingly bad.
KenoFischer··on On libunwind and dynamically generated code on x86-64
If you ever pick this up again, I'd recommend starting with llvm's libunwind library, which is an import of Apple's libunwind + support for other operating systems. As far as I can tell it's the only fully functional unwind library that's actually maintained at the moment. We discussed porting whatever functionality it is missing (probably linux support and the basic assembly profiling heuristics we added to our fork of Apple's libunwind) and just using it for julia on all platforms.
KenoFischer··on Time Machines
Let me try to phrase this a different way. Suppose the universe is a database, where (t,x,y,z) are the unique keys. For every key there is some value of "physical reality" at that point in spacetime. All the original statement is saying that if you want to generate such a database by starting with only some of the entries and applying the laws of physics, then in the presence of CTC, there are some very strict conditions on what those initial rows can be if you don't want to run into a situation where you'd want to assign two different values to the same key (e.g. by getting to the same point forwards in time and backwards in time).
KenoFischer··on Time Machines
Consistency here is a mathematical statement. Think about the cylinder example, and suppose you want to assign a number of each location. Then, naturally, going once around the cylinder puts you back at the same spot on the cylinder, so it needs to have the same number. At this point it is useful to remember that the numbers on the cylinder are fixed. We're already representing time as the compact dimension of the cylinder, so there's no extra time dimension.

Now, with this setup, let's imagine living in this 1D world, just sitting still, reading off numbers. Our life would be cyclic with period cm/((2pi r)s).

KenoFischer··on Time Machines
I'll have a go at defining it, term-by-term,

> smooth

Meaning no abrupt kinks (for a notion of abrupt which is probably not physically achievable, so don't worry to much about it) or jumps/discontinuities in the path of a particle, applying to both space and time

> timelike

roughly speaking, traveling slower than the speed of light

> future-directed

meaning going into the direction of causality

> with intersections

meeting itself again.

In other words, it's a path through space/time on which you experience your own local time as always running forward, but yet somehow you end up in the past.

Perhaps an analogy is in order. Suppose we were in 1+1 dimensions, with the time dimension being compact (closing in on itself). Visualize this as an infinite cylinder (the infinite axis being space). Now, let's impose a finite speed of light (for simplicity 1cm/s) and define the direction of causality, say counter-clockwise. Then, this spacetime has CTC, because we can find a path that, is smooth, future-directed (going around counter-clockwise) and timelike (going at angles < 45 degrees compared to the spacial axis). The simplest such would be a simple circle around the cylinder, but any curve that matches the above rules and intersects itself would be considered a CTC.

Now, as a final note " a crude way of bringing out the fact that the application of familiar local laws of relativistic physics to a spacetime background which contains CTCs typically requires that consistency constraints on initial data must be met in order for a local solution of the laws to be extendable to a global solution." means the following. When we usually do physics, we know what happens locally (e.g. ball get shot at wall, bounces back), so we simply have to figure out (or decide if we're doing a thought experiment) where everything is in the universe at some particular time, apply our rules everywhere and we essentially know what's happening everywhere, anywhere. This doesn't work any more in the presence of CTC. "Consistency conditions" basically just means that you have to chose your placement of objects in such a way that they don't cause any paradoxes (there's a more technical notion here, but that's essentially what's meant).

KenoFischer··on Julia: A Fast Language for Numerical Computing
Yes, I don't disagree. I don't know why the SIAM news editors didn't. I do hope my statement here makes clarifies the facts at least for people coming from here.
KenoFischer··on Julia: A Fast Language for Numerical Computing
Alan is one of the people who started the language (which is mentioned in the article). He is also one of the cofounders of Julia Computing (the company we formed to provide support and other services for the company). So from my biased viewpoint, the language is both truly amazing and Alan has a motive to promote the language, if only since he has a significant chunk of time and effort into making it a reality. I don't think it's fair to call it undisclosed however.
KenoFischer··on Julia: A Fast Language for Numerical Computing
HTTP site is up. It's hosted on Github Pages so HTTPS may not work.
KenoFischer··on Julia: A Fast Language for Numerical Computing
While numerical computing is the first target, that's definitely not the sole goal of the language, so if there's something simple that you feel is missing, I would be very curious to know what that is.
KenoFischer··on Julia: A Fast Language for Numerical Computing
It used to be a big problem. It's still not super fast (~1s or so), but we've been doing quite a bit of working at making it faster. If it's significantly slower than that, something else must be going on.
KenoFischer··on Julia: A Fast Language for Numerical Computing
> The literal syntax that you'd think would make an array of arrays concatenates the arrays instead. Sometimes. It depends on types. (Or have they fixed that now?)

Yup, we fixed that (deprecated in the latest release, will be switched over). It was super annoying.

There's been lots said on the 1-based indexing discussion, so I won't get into it here, but I can guarantee you that "open source Matlab" is not the ambition. Our ambition is to be the best possible programming language for technical computing and a great language for programming overall.

KenoFischer··on Best interview questions to spot ideal employees
And doing even more analysis, there were 87687 practicing doctors (not including dentists) in Poland in 2014, i.e. 2.3 per capita, which is close enough to that it could have been World Bank's data (which isn't available for 2014). Comparing those numbers to the US it just seems that Poland has a smaller dentist/doctor ration than the US.
KenoFischer··on Best interview questions to spot ideal employees
Actually I think it's just a difference in definitions. You were looking at practicing dentists, while the number I quoted was licensed dentists. If I'm interpreting the numbers correctly, there are only 13088 dentists actually practicing dentistry (I don't speak polish, so I'm not sure if the difference is explained in the report). I'm also not sure how the world bank data is derived.
KenoFischer··on Best interview questions to spot ideal employees
Exactly 40110 (in 2014, only those legally registered)

Source: http://stat.gov.pl/en/topics/health/health/health-and-health..., though you have to dig through a bit to find the figure.

KenoFischer··on Scientists struggle to stay grounded after possible gravitational wave signal
LIGO is very serious about doing blinded test runs, even going as far as writing the whole paper before they check whether the signal was real. There's only a couple of people in the entire collaboration who even know when the blinded test runs occur. In any case, this rumor has been going around for a couple of months, so if it's true, we should know soon. Between this and the LHC blip, I'm quite excited about the next couple of months.
KenoFischer··on Cxx.jl: C++ interpreter embedded in Julia
We'll it's been a pain to build so far, so I've been avoiding making it known more widely. That's about to change though.
KenoFischer··on Cxx.jl: C++ interpreter embedded in Julia
Oh, cool to see my little project on here, just in time for some lunch-time procrastination. Happy to answer any questions.
KenoFischer··on Cxx.jl: C++ interpreter embedded in Julia
It actually is a C++ interpreter ;), that functionality is just not advertised: http://imgur.com/miYn092
KenoFischer··on Physicists believe they can create matter from colliding photons
Yep, you'd be significantly better off just shining the laser behind you.
KenoFischer··on Physicists believe they can create matter from colliding photons
In theory if you crank up the energy high enough, you'll create a bunch of Higgs bosons as well, though of course as always, they'll immediately decay.
KenoFischer··on Physicists believe they can create matter from colliding photons
In theory all you have to do to get protons is crank up the energy.
KenoFischer··on Physicists believe they can create matter from colliding photons
Yes, in theory quarks and their anti-particles will be created just the same as electrons/positrons. However, a) it is a lot easier to detect electrons/positrons than it is to detect quarks and b) protons are significantly heavier, so you'd need quite a bit more energy. Really the best way to think about this is just a particle collider, but colliding photons instead of matter. You'll still get a whole sea of debris that you can sort through and find interesting physics, but just detecting electrons would be a good first step.
KenoFischer··on Physicists Hope to Be Wrong About the Higgs Boson
There isn't really a good notion of the size of an elementary particle. What's really being looked for is particles with certain higher (mass) energies. Now, why do we say that higher energies allow us to probe smaller scales? Well, at small scales(/short times), quantum mechanics allows these higher energy particles to play a role, even if there's not enough energy to create an actual particle. The smaller the scale, the more do high energy particles come into the picture. Thus cranking up the energy allows us to see these particles at a larger (detectable) scale than they would usually occur.
KenoFischer··on The Space Doctor’s Big Idea
It is worth pointing out that only about 10% of Mercury's parahelion shift is due to the effects of general relativity. The other 90% are due to gravitational influence of the other planets. It is really quite remarkable that astronomers of the 1800s made precise enough observations to be able to tell that the parahelion shift was not entirely explained by known objects in the solar system (since the effect is on the order of 10%, you have to know the mass and position of every other major object in the solar system to about that accuracy to be able to tell).
KenoFischer··on Julia Computing Granted $600k by Moore Foundation
In julia the rand function is more general in that it samples from a distribution, which you can pass as the first argument (defaulting to uniform on [0,1]). Since the first argument is an integer range, you get an integer value.
KenoFischer··on Xv6
There's also OS161, which is Harvard's version of this. It comes with a very simple MIPS emulator, on top of which the actual operating system is built. It's a lot of fun!
KenoFischer··on Julia Computing Granted $600k by Moore Foundation
That's definitely very high on the TODO list!
← PreviousPage 15 of 17Next →