ratsbane

The primary source: https://pages.nist.gov/800-63-3/sp800-63-3.html

I'm really surprised by these numbers. There isn't really any speed advantage or significant storage advantage to using tuples over lists. That really cuts down on the rationale for using tuples over lists, except for the "cultural difference" as discussed (and the ability to use lists as dictionary keys, which isn't something I think I've ever wanted to do)

Interesting! The previous example I posted was Python 2.7.12 on Ubuntu 16.04.01. Here is OSX 10.11.5 w/Python 2.7.10 (also inconclusive with respect to speed, but OSX was half the memory):

  mb:~ doug$ ./listtuple 
  The tuple's size is 40028 and the list's size is 40036
  Iterating through a 10,000 element list took 0.00113916397095 seconds.
  Iterating through a 10,000 element tuple took 0.00111603736877 seconds.
  mb:~ doug$ ./listtuple 
  The tuple's size is 40028 and the list's size is 40036
  Iterating through a 10,000 element list took 0.00126791000366 seconds.
  Iterating through a 10,000 element tuple took 0.00133585929871 seconds.
  mb:~ doug$ ./listtuple 
  The tuple's size is 40028 and the list's size is 40036
  Iterating through a 10,000 element list took 0.00114607810974 seconds.
  Iterating through a 10,000 element tuple took 0.00117492675781 seconds.

Another (admittedly trivial) experiment with speed and storage: Iterating through a 10,000 element list was about 2% slower than iterating through a 10,000 element tuple, and the list was 16 bytes bigger than the tuple:

  #!/usr/bin/python

  from sys import getsizeof
  from random import sample
  from time import time

  l = sample(xrange(10000), 10000)
  t = tuple(l)

  print "The tuple's size is " + str(getsizeof(t)) + " and the list's size is " + str(getsizeof(l))

  s = time()
  z = 0
  for x in l:
    z += x

  print "Iterating through a 10,000 element list took " + str(time() - s) + " seconds."

  s = time()
  z = 0
  for x in t:
    z += x

  print "Iterating through a 10,000 element tuple took " + str(time() - s) + " seconds."


  doug@supermicro:~$ ./x
  The tuple's size is 80056 and the list's size is 80072 
  Iterating through a 10,000 element list took 0.00589299201965 seconds.
  Iterating through a 10,000 element tuple took 0.00538778305054 seconds.

  doug@supermicro:~$ ./x
  The tuple's size is 80056 and the list's size is 80072
  Iterating through a 10,000 element list took 0.00553607940674 seconds.
  Iterating through a 10,000 element tuple took 0.00536203384399 seconds.

D'oh! Thank you, colanderman.

  >>> tuple = (94,)
  >>> getsizeof(tuple)
  64

Interesting article. I've always heard that about lists and tuples, but this article made me wonder how much less efficient with storage is a list.

The following trivial example shows that, at least for very small lists and tuples, except for the fixed overhead of 16 bytes for a tuple vs. 72 bytes for a list they're the same [edit: no it doesn't. I wasn't looking at the diff between a 1-element and 2-element tuple]:

Perhaps there's more to it?

  >>> from sys import getsizeof
  >>> list = [94]
  >>> getsizeof(list)
  80
  >>> list = [94, 37]
  >>> getsizeof(list)
  88
  >>> list = [94, 37, 21]
  >>> getsizeof(list)
  96
  >>> list = [94, 37, 21, 19]
  >>> getsizeof(list)
  104
  >>> tuple = (94)
  >>> getsizeof(tuple)
  24
  >>> tuple = (94, 37)
  >>> getsizeof(tuple)
  72
  >>> tuple = (94, 37, 21)
  >>> getsizeof(tuple)
  80
  >>> tuple = (94, 37, 21, 19)
  >>> getsizeof(tuple)
  88

There's nothing intrinsically expensive about a robot arm. It's few motors, sensors, metal, plastic, and something like a raspberry pi. With cheaper components and volume production a robot arm could cost less than a Domino's employee makes in a week.

I wonder how they power that. Propane?

Good find. An "extended major release" while they're onboarding 11 million new customers sounds painful.

The timing of this as the switchover of Costco cards from Amex to Citi is happening certainly suggests a scaling problem regardless of the "updating its systems" excuse.

I've heard Amex blamed for losing the Costco deal, but given Amex's superlative customer service and Citi's less-than-optimal reputation, maybe Costco will turn out to be the bigger loser.

Whether anyone "wants" or "needs" an AR-15 or any other type of firearm, the rest of us need to know that anyone around us with such a weapon has some basic competence in using it and is responsible enough to do so. I had to pass a written and and practical test and I have to show financial responsibility to operate a car. If you can do that with your AR-15 then I won't object to your having it.

I suspect you are a careful, conscientious, and responsible gun owner and would have no problem demonstrating this, but there are a lot of people who have guns who are neither careful, conscientious, nor responsible. I hope you won't mind having to demonstrate your competence so that society can stop the other people, the irresponsible ones.

And of course, you might say that even if we had such requirement to demonstrate responsibility, the irresponsible people would circumvent the system and have guns anyway. I'm sure that's true for a few, but it would remove many guns from irresponsible hands and it would give law enforcement a reason to arrest the others and get them off the streets.

I started saving file from my dashcam several months ago with that idea - to use the openalpr results as training data for a neural net, but the results I'm getting from openalpr aren't good enough to use without a lot of manual review.

phashes compute a hash on one image which you can then compare with the hashes for other images; butteraugli computes the difference between two images. If you're trying to find duplicates among a set of images, phashes will work in O(n log n) time and this will run in O(n^2) time, but with more accurate results.

I started training about 24 hours ago with a single i7 860 (2.8ghz) and it's still running, but I'm watching the loss going down so I think it might wrap up within a day or two.

The machine I'm using has an Nvidia GTX 970, but I failed to enable CUDA in the TensorFlow installation so it's not using it. When I noticed that I decided just to let it run and then, when it finishes, train it again with the GPU and compare.

I cloned this a few days ago and had some questions when I tried setting it up, and Matthew's been very helpful when I emailed him, so yes, seconded. Great project!

The openalpr project is pretty good, but it uses an older approach with lot of hand-coded logic instead of the convolutional neural net approach of this project. openalpr is faster than this, at least now, but I think the neural net has potential to be faster and more accurate.

I believe that St. Jude Children's Research Hospital does a pretty good job: http://www.charitynavigator.org/index.cfm?bay=search.summary...

https://www.stjude.org/

There are probably a lot more reserves out there that no one's found yet because the current reserves were sufficient for the demand. In the longer term there will be plenty.

In the short term, though, existing reserves are mostly in Argentina, Chile, and Bolivia. I wonder how difficult it would be for a cartel to manipulate pricing?

I know the guy behind this, David Lowe. He's been working on it for a long time. It represents a massive amount of research and work and will be useful to many people.

Congratulations to David, the team, and the NYPL!

A lot of models were not mentioned in that list: Audi A8, any Mercedes, etc. I wonder if they just weren't tested or if the hack doesn't work with them. (And if that's the case, what is different about those models?)

You might be happy to hear that Jesus College at Cambridge just decided to remove the Benin Bronze they've had since 1897 and that it might go back to Nigeria: http://www.telegraph.co.uk/education/12188018/Cambridge-coll...

I took that class on the Stanford campus in the Summer of 2011. Keith was one of the most engaging instructors I've ever had. His enthusiasm and knowledge were boundless.

The course was video-recorded and I downloaded and saved most of the lectures (and I gave Keith a DVD with a copy of them.) I still have them, but I don't know if it's okay to put them on YouTube or not.

I tried Let's Encrypt for the first time a week or two ago and was really impressed. This is the way to do it. Well done!

Since it's a skid-steer (steering consists of slowing the rotation of wheels on side you want to turn towards) the whole setup is probably different. I'm guessing it has two hydraulic motors, one for the left side and one for the right side, powered by a hydraulic pump driven by the engine.

Donald Knuth's annual Christmas lecture at Stanford was just released on YouTube a few days ago. It's about comma-free codes, a similar idea to this bug: https://www.youtube.com/watch?v=48iJx8FVuis

Surely that can't be for real. I saw it a few days ago and immediately thought it was a parody... but...

The FAA has created a lot of FUD by not drawing clear lines on what they don't control.

I'm a private pilot with rotorcraft rating as well as frequently a passenger and there's nothing I'd rather not have happen than a DJI Phantom through the windshield, and I don't think the FAA's actions are helping. By attempting to ban actions that are clearly not unsafe they're promoting disrespect for the law and crippling development of the drone industry in the U.S.

What they need to say: no UAVs above 400' AGL or within a mile of airports, or whenever aircraft are operating in the vicinity (without mitigating controls). The commercial/non-commercial distinction is irrelevant.

Also not available in Tennessee, Arkansas, North Carolina, and Virginia. (Those were the only states I checked)

Was the server actually at her house or was the domain name just registered with that as an address? Neither option seems even slightly damning.

This illustrates something that might be a significant future problem for law enforcement, etc. - how to enforce anti-drone laws. Perhaps this will lead to an anti-drone technology industry with drone-tracking equipment and drone-capture drones. With radio-controlled devices, direction-finding equipment could work but with autonomous drones...?