fluid-work IRC Logs-2013-01-25

[09:15:39 CST(-0600)] <jessm> lest we all make assumptions about the connectivity of the "modern" world: http://www.globalpost.com/dispatch/news/regions/americas/cuba/130124/cuban-undersea-internet-cable-web-connection

[16:06:02 CST(-0600)] <colinclark> Bosmon: Hey, you merged my pull request!

[16:06:05 CST(-0600)] <colinclark> thanks so much!

[16:06:18 CST(-0600)] <Bosmon> Thanks, colinclark

[16:06:27 CST(-0600)] <colinclark> Justin_o asked me to ask you all about my ancient build-related pull request

[16:06:36 CST(-0600)] <colinclark> I don't even know if it will apply cleanly anymore

[16:06:45 CST(-0600)] <colinclark> but avtar is working on migrating our build infrastructure to Jenkins

[16:06:49 CST(-0600)] <Bosmon> I'm just writing a small note to the list now about the new testing changes

[16:06:55 CST(-0600)] <Bosmon> I'm certain it won't apply cleanly, colinclark

[16:07:13 CST(-0600)] <Bosmon> I was discovering last night that "magical" git merging isn't really as magical as one might expect

[16:07:26 CST(-0600)] <Bosmon> It made a total mess of trying to merge my testing changes against the new IoC framework

[16:07:26 CST(-0600)] <colinclark> to go with a new TestSwarm instance and cloud-based VMs (Mac and Windows for each browser)for automatically testing with TestSwarm

[16:07:42 CST(-0600)] <colinclark> Bosmon: yes, I've noticed that it's not so magical, too

[16:07:49 CST(-0600)] <colinclark> but I've never had a horrible mess, at least

[16:08:09 CST(-0600)] <colinclark> anyway, I imagine avtar will appreciate not having to install Maven and the Ant-Maven plugin if we can avoid it

[16:08:19 CST(-0600)] <Bosmon> It really was a horrible mess and hadn't even "conserved" the existence of functions across the branches

[16:08:31 CST(-0600)] <colinclark> that's awful

[16:08:39 CST(-0600)] <Bosmon> colinclark - why would he need to install those plugins?

[16:09:04 CST(-0600)] <colinclark> because our build process still sadly creates a war file

[16:09:05 CST(-0600)] <avtar> because avtar likes to install all the plugins

[16:09:14 CST(-0600)] <colinclark> That's what I removed in my old pull request

[16:09:16 CST(-0600)] <colinclark> let me dig it up

[16:09:32 CST(-0600)] <colinclark> ah!

[16:09:36 CST(-0600)] <colinclark> I never actually made the request

[16:09:40 CST(-0600)] <Bosmon> I see

[16:09:41 CST(-0600)] <colinclark> avtar can now laugh at me

[16:09:44 CST(-0600)] <Bosmon> Yes, I couldn't see it being open

[16:09:46 CST(-0600)] <colinclark> I sent an email to the list

[16:09:49 CST(-0600)] <colinclark> and no one responded

[16:09:53 CST(-0600)] <Bosmon> I knoew that your branch is there

[16:10:21 CST(-0600)] <Bosmon> I am certainly in favour of your branch

[16:10:40 CST(-0600)] <Bosmon> But am feeling a bit "tuckered out" as sgithens might say in dealing with widespread and mechanical issues which touch the whole codebase : P

[16:11:43 CST(-0600)] <colinclark> http://old.nabble.com/Removing-WAR-file-builds-from-Infusion-1.5-to34293202.html#a34313953

[16:11:51 CST(-0600)] <colinclark> Bosmon: Yes, don't worry about it now

[16:12:03 CST(-0600)] <colinclark> Would you be consider reviewing it some time next week?

[16:12:13 CST(-0600)] <colinclark> avtar has some work ahead of him, I think, before he'll get to Jenkins and TestSwarm

[16:12:19 CST(-0600)] <colinclark> but I wont' speak for him

[16:12:40 CST(-0600)] <Bosmon> colinclark - I'd love to

[16:12:40 CST(-0600)] <avtar> that sounds about right

[16:12:49 CST(-0600)] <Bosmon> Would you be able to update it to work with current trunk?

[16:12:54 CST(-0600)] <colinclark> Bosmon: for sure!

[16:12:56 CST(-0600)] <colinclark>

[16:12:57 CST(-0600)] <Bosmon> Great

[16:13:15 CST(-0600)] <avtar> i also would like to spend some time taking a look at how other javascript projects might be using jenkins

[16:13:20 CST(-0600)] <colinclark> that's a good idea

[16:13:34 CST(-0600)] <colinclark> I would REALLY like to rewrite our build system with Grunt

[16:13:43 CST(-0600)] <colinclark> and I really only think it would take a couple of days

[16:13:49 CST(-0600)] <colinclark> but focused days are so rare for me

[16:13:49 CST(-0600)] <Bosmon> Silly name

[16:13:54 CST(-0600)] <colinclark> Grunt?

[16:13:56 CST(-0600)] <Bosmon> yes

[16:14:11 CST(-0600)] <colinclark> I've heard worse

[16:14:14 CST(-0600)] <colinclark> "Flocking"

[16:14:19 CST(-0600)] <Bosmon> "alpha nerd"? : P

[16:14:29 CST(-0600)] <colinclark> lol

[16:14:38 CST(-0600)] <colinclark> thealphanerd: Bosmon's trash talking you!

[16:14:48 CST(-0600)] <colinclark>

[16:14:52 CST(-0600)] <thealphanerd> WHAAA

[16:14:54 CST(-0600)] <thealphanerd>

[16:15:53 CST(-0600)] <thealphanerd> Bosmon: I thought we were bros bro

[16:16:00 CST(-0600)] <colinclark> what's new thealphanerd?

[16:16:02 CST(-0600)] <Bosmon> Sure we are

[16:16:08 CST(-0600)] <Bosmon> That's why I can tell you about your silly name

[16:16:11 CST(-0600)] <thealphanerd> trying to do me some dsp homework

[16:16:13 CST(-0600)] <colinclark> hahaha

[16:16:20 CST(-0600)] <colinclark> what's in the homework

[16:16:21 CST(-0600)] <colinclark> ?

[16:16:22 CST(-0600)] <thealphanerd> trying to figureo ut what the heck the question wants me to do

[16:16:41 CST(-0600)] <thealphanerd> http://cl.ly/45003w0n0936

[16:16:43 CST(-0600)] <thealphanerd> that's the homework

[16:16:54 CST(-0600)] <thealphanerd> http://cl.ly/1V1C1K2f0d3U

[16:16:57 CST(-0600)] <thealphanerd> this is where I am at so far

[16:17:22 CST(-0600)] <colinclark> which question are you stuck on?

[16:17:47 CST(-0600)] <Bosmon> Yay

[16:17:50 CST(-0600)] <Bosmon> You have found some POLES

[16:18:04 CST(-0600)] <thealphanerd> 2 g

[16:18:16 CST(-0600)] <thealphanerd> Digitize the state-space ODE using the forward difference method to arrive

[16:18:16 CST(-0600)] <thealphanerd> at an explicit finite-difference scheme that computes a sample of displacement

[16:18:18 CST(-0600)] <thealphanerd> xm(nT) as a function of previous displacement samples and driving-force samples.

[16:18:18 CST(-0600)] <thealphanerd> Describe how this finite-difference scheme might be improved.

[16:18:39 CST(-0600)] <thealphanerd> I think I have the right answer

[16:18:44 CST(-0600)] <thealphanerd> but not sure how it could be improved

[16:19:07 CST(-0600)] <Bosmon> It could be improved by picking a better discretisation, I imagine

[16:19:19 CST(-0600)] <Bosmon> For example, Euler methods don't guarantee even basic things like energy conservation

[16:19:51 CST(-0600)] <thealphanerd> I thinkk I need to read on what exactly a finite difference scheme is

[16:20:11 CST(-0600)] <Bosmon> Well, it's one of exactly what you've got there - a recurrence relation indexed by integers

[16:20:27 CST(-0600)] <Bosmon> Which in this case, you hope somehow "approximates" a sampling of a function described by an analogous differential equation

[16:20:28 CST(-0600)] <thealphanerd> I had some hand holding to get to where I am right now

[16:20:55 CST(-0600)] <Bosmon> There will always be some kind of error in this process of discretization since you could never hope to completely faithfully represent the continuous differential operators

[16:21:13 CST(-0600)] <Bosmon> But you can pick different strategies for discretisation that are more or less faithful in different ways

[16:21:33 CST(-0600)] <thealphanerd> so my state space machine

[16:21:35 CST(-0600)] <Bosmon> Of which the absolute simplest is the Euler scheme.... you just convert every differential operator into the equivalent difference operator

[16:21:42 CST(-0600)] <thealphanerd> state space model

[16:21:50 CST(-0600)] <thealphanerd> is a representation of the continious system

[16:21:55 CST(-0600)] <Bosmon> yes

[16:22:24 CST(-0600)] <thealphanerd> so I have an ODE

[16:22:32 CST(-0600)] <thealphanerd> ordinary difference equation

[16:22:50 CST(-0600)] <thealphanerd> is the difference between an ODE and an FDE that FDE is discrete?

[16:22:57 CST(-0600)] <Bosmon> I'm trying to find some links on different discretizations.... you would think that there would be as many of these as there are ODE solvers

[16:23:00 CST(-0600)] <Bosmon> thealphanerd - that's right

[16:23:45 CST(-0600)] <Bosmon> There may be some particular scheme that your lecturer has in mind..... did he give you any notes?

[16:23:59 CST(-0600)] <thealphanerd> https://ccrma.stanford.edu/~jos/intro420/Schedule.html

[16:24:04 CST(-0600)] <thealphanerd> this is the course outline so far

[16:24:13 CST(-0600)] <thealphanerd> https://ccrma.stanford.edu/~jos/pasp/

[16:24:15 CST(-0600)] <thealphanerd> the text

[16:24:21 CST(-0600)] <Bosmon> thanks

[16:24:30 CST(-0600)] <thealphanerd> and on that scedule is the info for all the lectures

[16:24:45 CST(-0600)] <thealphanerd> https://ccrma.stanford.edu/~jos/DigitizingNewton/

[16:24:49 CST(-0600)] <thealphanerd> I should probably go through this

[16:25:30 CST(-0600)] <Bosmon> So yes, he's probably thinking of this as a simple improvement

[16:25:32 CST(-0600)] <Bosmon> https://ccrma.stanford.edu/~jos/DigitizingNewton/Centered_Finite_Difference.html

[16:25:32 CST(-0600)] <thealphanerd> I just want to make sure I am understanding this stuff before I move forward

[16:25:47 CST(-0600)] <Bosmon> By using centred differences you can cancel out some kinds of error to first order

[16:26:17 CST(-0600)] <thealphanerd> ahhh

[16:26:25 CST(-0600)] <thealphanerd> is it like an average of the forward / backward

[16:26:28 CST(-0600)] <thealphanerd> ?

[16:26:35 CST(-0600)] <Bosmon> Yes, something like that

[16:26:43 CST(-0600)] <thealphanerd> Bosmon: did you study EE?

[16:27:03 CST(-0600)] <Bosmon> Basically you can pick from a huge variety of discretisations, which an analysis can show have various beneficial properties in representing the original ODE

[16:27:45 CST(-0600)] <Bosmon> This links onto a very wide literature on ODE solvers which you can dip into here : http://en.wikipedia.org/wiki/Numerical_ordinary_differential_equations

[16:27:55 CST(-0600)] <Bosmon> I did maths, originally

[16:28:06 CST(-0600)] <Bosmon> But I supervised the Engineering Mathematics course for a while : P

[16:28:33 CST(-0600)] <thealphanerd> This stuff is tough for me

[16:28:36 CST(-0600)] <Bosmon> So you can see the forward and backward Euler methods referred to on this page

[16:28:42 CST(-0600)] <colinclark> thealphanerd: He's British, so it's plural there

[16:28:43 CST(-0600)] <Bosmon> Which are ones that your lecturer gave first

[16:28:52 CST(-0600)] <thealphanerd> 4 months ago I learned what j / i is

[16:29:03 CST(-0600)] <thealphanerd> and just last week I learned about derrivatives and integrals

[16:29:10 CST(-0600)] <thealphanerd> never really done any of this stuff

[16:29:12 CST(-0600)] <Bosmon> thealphanerd - a tough learning curve!

[16:29:17 CST(-0600)] <Bosmon> But it's all brilliant stuff

[16:29:19 CST(-0600)] <thealphanerd> so I kind of feel like I swimming in the deep end

[16:29:25 CST(-0600)] <Bosmon> This is what everything is made of

[16:29:45 CST(-0600)] <Bosmon> Or rather, what one of my lecturers would describe as "what every gentleman should know"

[16:29:47 CST(-0600)] * thealphanerd wants to get it all but struggles with the algebra

[16:30:07 CST(-0600)] <thealphanerd> but I'm not going to stop.. I'm just going to keep taking dsp classes as long as I am here

[16:30:19 CST(-0600)] <thealphanerd> no way to learn it other than by doing it

[16:30:33 CST(-0600)] <Bosmon> thealphanerd - good job!

[16:30:39 CST(-0600)] <thealphanerd> yay!

[16:30:57 CST(-0600)] <Bosmon> Do come and ask if you get hung up on anything

[16:31:00 CST(-0600)] <Bosmon> http://ocw.usu.edu/civil_and_environmental_engineering/numerical_methods_in_civil_engineering/ODEsMatlab.pdf

[16:31:15 CST(-0600)] <Bosmon> Here's a little guide showing the 3 different discretization schemes your lecturer has shown

[16:31:19 CST(-0600)] <Bosmon> It puts them in context with a few examples

[16:31:28 CST(-0600)] <thealphanerd> oh cool

[16:31:29 CST(-0600)] <thealphanerd> !

[16:31:32 CST(-0600)] <Bosmon> But really these things can become arbitrarily complex, depending on how good you need them to be

[16:31:36 CST(-0600)] <thealphanerd> the lecturer is Julius Orion Smith fwiw

[16:31:44 CST(-0600)] <thealphanerd> guy is crazy brilliant

[16:31:52 CST(-0600)] <colinclark> he is indeed

[16:32:08 CST(-0600)] <thealphanerd> colinclark: are you going to be able to visit?

[16:32:14 CST(-0600)] <colinclark> I still don't know

[16:32:17 CST(-0600)] <Bosmon> "fully grown up" people who need to do do things like integrating planetary systems will use very advanced schemes like this: http://en.wikipedia.org/wiki/Symplectic_integrator

[16:32:34 CST(-0600)] <colinclark> I talked to Adam Tindale about it--he said he'd send a note to Julius about it if I could make the time

[16:32:38 CST(-0600)] <Bosmon> They've recently used these to discover entirely new ways of navigating between satellite orbits!

[16:32:43 CST(-0600)] <colinclark> It'd have to be May, I think

[16:32:54 CST(-0600)] <colinclark> Bosmon: I hate being a kid

[16:33:01 CST(-0600)] <Bosmon> But this relates to the issue of "conserved quantities" as I mentioned at the outset

[16:33:13 CST(-0600)] <colinclark> Somehow I vaguely inuit things from reading Julius' book

[16:33:14 CST(-0600)] <thealphanerd> colinclark: classes are still going in may

[16:33:15 CST(-0600)] <Bosmon> Sometimes there are really important aspects of the system you need conserved, even if you get the details slightly wrong

[16:33:18 CST(-0600)] <thealphanerd> and henke will still be here

[16:33:19 CST(-0600)] <colinclark> but it's very slow going

[16:33:24 CST(-0600)] <Bosmon> And energy conservation is one of the most important of these

[16:33:37 CST(-0600)] <thealphanerd> colinclark: if you want me to speed it up let me know. I see julius almost every day

[16:33:42 CST(-0600)] <thealphanerd> we're even facebook friends

[16:33:52 CST(-0600)] <colinclark> Hopefully by then I will have my Flocking "boxes" working fully

[16:34:33 CST(-0600)] <thealphanerd> !!!

[16:34:39 CST(-0600)] <thealphanerd> I can maybe help you with it this quarter

[16:34:44 CST(-0600)] <thealphanerd> I have to do independent research

[16:39:53 CST(-0600)] <colinclark> that'd be cool

[16:42:06 CST(-0600)] <thealphanerd> what work needs to be done?

[16:42:18 CST(-0600)] <thealphanerd> I was going to start by doing a round up of dsp technologies inthe browser

[16:49:51 CST(-0600)] <colinclark> thealphanerd: I think mostly stuff just needs to be done

[16:49:58 CST(-0600)] <colinclark> performance optimizing

[16:50:08 CST(-0600)] <colinclark> some kind of OSC binding and MIDI

[16:50:20 CST(-0600)] <thealphanerd> ohhh I could defeinitely help with osc / midi

[16:50:25 CST(-0600)] <thealphanerd> maybe even do some monome stuffs

[16:50:28 CST(-0600)] <colinclark> that would be very cool

[16:50:38 CST(-0600)] <thealphanerd> let me talk to julius and see if I can do it for my project

[16:55:54 CST(-0600)] <colinclark> that'd be really neat

[16:56:52 CST(-0600)] <thealphanerd> !!!

[17:25:39 CST(-0600)] <Bosmon> thealphanerd - so in the case of the centred difference system, the advantage of that discretisation is that it will not produce a 1st-order time advance or time delay, as either of the Euler schemes would