Hacker News Reader

zevetson Larry Page is pouring millions into flying cars

iirc a v-22 is below 5:1 and somewhere around 3:1

The new DEP designs will do better, but exceeding 10:1 seems unlikely

my understanding is that maneuverability is thought of in Gs - so even though the turn radius is enormous, the acceleration is high, and it's the acceleration that prevents defensive interceptors from getting hits

zevetson Ask HN: Who wants to be hired? (October 2016)

  Location: Washington DC
  Remote: No
  Willing to relocate: Yes
  Technologies: CUDA, MATLAB,C,C++,Python
  Résumé/CV: Soon to be Ph.D. grad in aerospace eng. Research in fluids, structures, magnetics, optimization and a bit of robotics. Strong self-taught programming skills.
  Email: steve g sherman at gmail

zevetson Shape-programmable magnetic soft matter

This paper looks a lot like a materials hype paper. The mathematics for the design process are interesting, but the material/actuator itself looks pretty poor.

The actuation forces are incredibly low - if i've eyeballed correctly - around 10 micro newtons and it requires a large bulky magnetic field setup, only shown in the supplementary section.

zevetson Now is a really good time to buy a helicopter

Helicopters are more industrial equipment than luxury goods. Most sales are military, medevac/police or offshore oil.

zevetson How Much Is a Good Teacher Worth? (2011)

If this was the case, then why do private school teachers get paid less? Private schools are the free-market and are ununionized.

In case you're curious, the reason is that teaching at a private school is easier, so people who love teaching are willing to take the pay cut.

zevetson The End of Numeric Error: An interview with John L. Gustafson

But will unums help? No matter how an ill conditioned matrix is represented, it will still be ill conditioned, and we'll still be iterating, unless I've missed something in this article.

I'm skeptical that changing how we represent the numbers will help - when the physics are icky, they're icky.

zevetson Doom Creator John Carmack Honoured with Bafta

I'm fascinated that Carmack thinks Quake 3 is overlooked. At least among the people I know, Quake 3 is considered the best multiplayer arena shooter of all time.

zevetson The Nvidia DGX-1 Deep Learning Supercomputer in a Box

Any finite element/volume problem. Anything integrated.

zevetson Gallium nitride is about to disrupt silicon?

Military radars and other high power electronics

zevetson Agricultural – fastest growing robotic sector

I think VCs are scared by the sensible business plan and the larger capital costs. Drones also get a pass from scrutiny as VCs get excited about investing in the platform, which allows VCs to project their 100x dreams.

I would guess your biggest challenge is sales - convincing farmers that they need this. Is this true? Also - out of curiosity, what do you fly?

zevetson SunEdison at risk of bankruptcy, unit says; shares plummet 60 percent

What's the risk of default if the subsidy structure for solar goes kaput ie nevada?

zevetson Agricultural – fastest growing robotic sector

That's neat to see - I had always expected large vehicles to dominate the surveillance space. Will you guys go unmanned when, 10 years from now, the FAA lets civilians fly Predator like UAVS? Also, what's your perspective on the commercial smallsat imagery market? I've long been of the view that large UAVs will wipe those guys out.

zevetson Boom (YC W16) – Supersonic Passenger Airplanes

Since the founder is in the thread, why did they target cruise > Mach 2 and a 40 passenger size? The business jet segment makes a lot of sense for high speed flight, but this seems like more of a stretch.

zevetson Adam

I have not. That's super neat and I'm glad people are making progress! Is there anything happening on the animation side which will push past mocapped movements?

zevetson Adam

I'm claiming that at this level hardware power, the hardware power isn't the limiting factor in realism, the animations are.

Realistic visuals needs realistic walk cycle. But since realistic walking isn't a perfect stable cycle (it is disturbed by head/hand motions, surface variations, etc), there can be no realistic walk cycle because real walking isn't a cycle. AFAIK current videogame animation all involves walk cycles. Realism at this level of detail needs some new form of animation, not the love and care of a person that is so present in this demo.

zevetson Adam

That sounds like the step forward these things need. VR to really get a good look at your stuff, and the ability to mocap while sitting at your pc. Cool stuff.

zevetson Adam

A two minute film is not a videogame. A show reel requires one good animation to work in one specific situation. A videogame needs an animation that works generally. Also, making a good animation is hard, and artists deserve to be paid. Think of movies, where there are still movies today which have unrealistic CG, despite computation being far from a bottleneck for a feature film (Legolas on the elephant in LOTR: Two Towers) - it's not the technical challenges, it's the artistic challenges.

Right now, the vast majority of games have characters with fixed walk cycles that are used no matter the terrain. Realistic CG needs to have a walk cycle that captures the subtle changes in gait that corresponds to a given surface. I know people have been doing research on adaptive walk cycle, but afaik it has yet to hit production games.

For generating art, there is hope, as procedurally generated games look fantastic (No Man's Sky), but have yet to expand beyond sci-fi games, or into games with story and specific art styles.

Maybe reusing assets is the way forward, but I'm skeptical. Reusing the visuals just means more army guys fighting in sandy deserts crouching behind crates. Maybe the Storm Trooper models for battlefront really are as good as they get. But even then, I think aesthetically realism can lead to a dead end. The most visually impressive game I've played, other than Star Wars Battlefront is The Witness, which is simple in the CG sense, but has some really tremendous visual aesthetic moments that are something I've never seen done in a game. For me, the stylized look and the artistic/game opportunities that enabled were far more exciting than Battlefront's perfect rocks.

zevetson Adam

Even then, I'm unconvinced that the ability to do realistic graphics in real time will actually manifest itself in vediogames/VR experiences we can interact with. So much of what makes this look good is the excellent art direction, and as you stated, excellent textures and animations of an artificial thing. But animation students have been producing similar quality visuals for their show-reels for ages. All that has changed is the ability to render faster, but the creation of this content is still being done by the same processes.

For videogames/vr to break the photorealistic barrier, there needs to be some order of magnitude reduction in art development costs for these experiences to be affordable. Not all videogames can be star wars battlefront, where probably $150M ($330M overall cost to develop and market [1]) was spent making the best damn textures videogames have ever seen, but produced a simple, limited game.

Photorealism is a dead end for videogames unless art costs come down.

[1] http://fortune.com/2015/12/30/star-wars-video-game-sales/

zevetson Why huge gains in drug discovery tech have led to longer drug development times

The article's conclusion would indicate that you would do use in-silico biology to 'discover' existing drugs. Then once computer methods can reliably discover, say both penicillin and tetracycline, you'd push forward.

Funding this, however, would never come from industry. You'd need to convince NIH that spending many years rediscovering old drugs is the magic leap forward we need.

As someone ignorant of the field - has this been done? What is the state of validation in computer biology?

zevetson Why huge gains in drug discovery tech have led to longer drug development times

I didn't get that at all from either the interview or the paper.

From the paper, the conclusions seem to be: Decision theory creates a logically grounded explanation as to why high predictive value experiments are being done less often. To resolve this, research needs to be done in measuring the effectiveness of drug discovery models.

Nowhere in the article do they call for more math - the final conclusion of the paper is a call for more experiments validating existing drug models.

zevetson Ten lessons I wish I had learned before teaching differential equations (1997) [pdf]

I recently saw them in a grad eng class, but I agree with the article - from what I saw there is no need to give them the math professor treatment. You can use them as a piece of trivia - ie pde of type x has this set of basis functions - now apply the principles of basis functions to solve your your problem.

zevetson The unsung heroes of scientific software

I can easily imagine such an article, because it happens all the damn time.

"Using the argument of [1] and [2] it can be shown, with brief step x, it can be shown that this theorem holds in xxxx" and then moving onto the conclusions, is exactly the same as "Using the software developed in [3] to solve Eq. 1, we show that our result is statistically significant."

If the claim is probable, it will and does get through review.

zevetson Thomas Kuhn changed the way the world looked at science (2012)

I read Kuhn in my second year of grad school and his bit on the structure of a scientific paper I found very helpful. His section on what goes in an introduction to was supremely helpful in demonstrating how to frame your work so that it finds acceptance. After reading that portion, I observed that I was getting much nicer peer reviews, and I've since recommended it to several people.

zevetson Show HN: The Autoprotocol Language Standard for Biology

This is neat, but you guys have failed on the basic human motivation. A python file isn't something I can take into the lab with me to reproduce an experiment. Does this spit out a useful human readable set of instruction?

Is there a sample output spec, compared with the norm of human readable input? As is, this seems like a solution for computers, not folks in the lab.

zevetson Compiling Julia for NVIDIA GPUs

Have you tried thrust? For a limited but very useful set of operations, it is nearly that painless, as long as you don't mind a generous helping of C++ template boilerplate.

http://thrust.github.io/

For single GPU stuff, I find myself writing the performance critical kernels in cuda, and using thrust for nearly all the easy stuff.

zevetson On Immigration, Engineers Simply Don’t Trust VCs

Does your employer use the lots of startupland hiring phrases, like "we're looking for passionate rockstars who believe in our mission to change the world through payroll software"?

A lot of the phrases in vogue right now seem to indicate companies that want employees that work 80 hour weeks and pay entry level salaries. Have you tried "competitive compensation" or some other non-numerical indicator that you guys pay well or phrases that indicate you guys have a professional vs. passionate culture?

zevetson The U.S. Navy Tests Its Ships in This Indoor Ocean

For the type of stuff they do at Carderock, I'd say experiments are way better. Maneuver dynamics is really, really hard to do, as the time scales of the problem are long (ships turn on the order of a minute), something very challenging for CFD to accurately represent. Also, accurate free surface liquid dynamics is really damn hard to do, and I'd be skeptical of of most model's accuracy. I certainly wouldn't make billion dollar decisions on it.

In aerodynamics, which has had much more time and money invested, CFD can almost completely replace the wind tunnel, but only for aerodynamics problems we understand (ie wings, fairings, nacelles, turbine blades, backward facing steps, etc). As soon as you leave the validated/understood world, CFD becomes a tool used to augment the designer's intuitions, with the wind tunnel providing an essential ground truth.

Fluid dynamics is the business of predicting complex nonlinearities to a few %, and imo, complex nonlinear models are only useful when validated. It's telling that when they debated building the DDG1000 they built a 1/4 scale ship:

http://upload.wikimedia.org/wikipedia/commons/4/4a/Advanced_...

zevetson Earthquake early-warning system gave 10-second alert before Napa quake felt

I attended a talk by one of the developers of this system, and since earthquake warnings are so short (tens of seconds), it's unlikely that traditional alert mechanisms can be delivered to the general public. As such, the primary consumers of the device will be places like nuclear power plants and major industrial operators who are willing to install special alarms.

The other example he gave was giving doctors performing surgery time to step away from the patient, so that they dont turn things into a real life game of surgeon simulator, an example I personally find to be extremely horrifying and nightmare inducing.

zevetson World’s First Carbon Fiber 3D Printer Announced

Good catch on them showing E vs strain instead of stress/strain. Very misleading, as it clearly got me!