One thing I would like to point out, and I hope as many people as possible hear me, is that there is a large barrier between the theoretical knowledge covered in text books and actual implementation. For example, I might be well versed in signal processing, but I may not know how to go about implementing it in a computer in an efficient manner. Therefore, a question that sounds like mere ignorance of basic concepts may, in fact, be simply a matter of picking the brains of experienced, practicing DSP engineers. One can read all of the text books they want, but when it comes to implementation, its best to ask somebody with experience about what is best---even if its regarding a very "simple" topic.
Group Culture Question
Started by ●February 17, 2009
Reply by ●February 19, 20092009-02-19
Reply by ●February 19, 20092009-02-19
On 19 Feb, 06:49, "joncox" <jon...@mit.edu> wrote:> One thing I would like to point out, and I hope as many people as possible > hear me, is that there is a large barrier between the theoretical knowledge > covered in text books and actual implementation.Sure. When did you ever see a DSP textbook that actually discusses applications or shows examples of data? I know of only one, Robinson & Treitel: "Geophysical Signal Analysis" (1980) but it is first of all old, and also targets a specific application. And it doesn't cover the data anlysis questions in all too much depth (no pun intended).> For example, I might be well versed in signal processing, but I may not > know how to go about implementing it in a computer in an efficient manner. > > Therefore, a question that sounds like mere ignorance of basic concepts > may, in fact, be simply a matter of picking the brains of experienced, > practicing DSP engineers. One can read all of the text books they want, but > when it comes to implementation, its best to ask somebody with experience > about what is best---even if its regarding a very "simple" topic.This is one of the tricky issues. I've always concentrated on concepts rather than details, when it comes to implementation. I always preferred to get a matlab protoytype up and running fast, to see if the idea works, rather than have a very efficient implementation that might not work. And I have relied on experts to help me if and when efficiency becomes a problem. The thing is that these days, it is not about getting one cool algorithm up and working; it's about integrating lots of small components into big systems. So I often need to implement some variation of some scheme here, or adapt a method there. In existsing systems. Which means I have to implement lots of stuff more or less from scratch, for legal reasons, portability reasons and system integration reasons. For instance, I know that one needs to handle numerical absolute and relative tolerances, but how do you actually do it? I also know that simple rearrangements of certain formulas can improve numerical accuracy, but where does one learn about these things? Not easy questions to answer without guidance. So I need help from numerics experts. I have usually found myself in one of two (percieved) situations: 1) The expert is perfectly willing to help, but is somehow unable. 2) The expert clams up with the interesting details In case 1) the experts, who usually are seniors with decades of experience in low-level numerical implementations, are unable to see the problem from my point of view. They take certain basic knowledge for granted, much like I take Euler's formulas for granted, and just don't understand what level of detail I needed help with. Or in case 2) they are perfectly aware of both what help I need, but choose to clam up to protect their own turf. Or to protect business interests. Which might be the sensible thing to do, from their POV, as these types of questions can make or break a product or project. As somebody commented, 'why help the competition?' The only remedy is to have students implement their own stuff a least in intro classes (matlab or scilab is perfect for such use), and not use packages or libraries until they become tools in later classes. This way students get hands-on experience with implementations and might develop an intuition around computational and numerical issues and trade-offs. Rune
Reply by ●February 26, 20092009-02-26
On Tue, 17 Feb 2009 23:07:03 -0700, Eric Jacobsen <eric.jacobsen@ieee.org> wrote:>On Tue, 17 Feb 2009 09:55:28 -0800 (PST), HardySpicer ><gyansorova@gmail.com> wrote: > >>On Feb 18, 3:24�am, "joncox" <jon...@mit.edu> wrote: >>> I have a question regarding the culture of this group. I noticed there are >>> some beginners asking simple questions with only the desire to get pointed >>> in the right direction to start their learning process. However, typically >>> the response they get is rude, sarcastic and unhelpful. >>> >>> What is with this? How can we as engineers be anti-learning/teaching? Is >>> this the legal profession? The attitude seems to be "if you don't know, >>> leave it to the professionals." I don't get it. This isn't engineering as I >>> know it... >> >>Piss off!I learned quite quickly how to ask a question on Usenet. One of the first things I learned was how to use Deja News, er, "Google Groups" to find threads where the topic had already been discused - this often not only answered my question, but answered questions I didn't know to ask, and provided a wealth of learning. It also helped me not ask so many questions, which also greatly helped to cut down on the snarky responses I got. I swear there's a webpage that tells "how to ask a question on Usenet."> >lol > >In addition to the numerous reasonable replies already given, I'll add >that sometimes what may seem to be an inappropriate response to a new >poster is really a conditioned response to a frequent poster who has >repeatedly failed to demonstrate an ability to grasp basic concepts >already explained. > >Sometimes it's just misunderstandings due to poor English language >skills on the part of the poster,And sometimes it's technical misunderstandings. I recall a poster explicitly saying he only needed calculations to four significant figures, but others in the thread were confused about the accuracy the OP wanted because he copy/pasted a small data set of values, each to 15 or 20 significant figures, and so they were thinking based on that. It's coming back to me now - that poster started two or three related threads, it seemed he wasn't sure what he wanted or how to ask about it, and it was getting frustrating. This is where stuff like the following happens:>and sometimes (although I'd say >pretty rarely) there may be a genuinely rude response that wasn't >provoked. That's not to justify any rude or brusk responses, just >offer some explanation. This is, after all, unmoderated intartubes, >and as such comp.dsp is actually pretty civilized.Over the years, one common response to a new poster becoming upset over a rude response is "Welcome to Usenet." It's quite different from web-based forums. They aren't just moderated, they're often heavily moderated, being power trip for the owner/moderator - often saying the "wrong thing" or saying it the wrong way can get you banned as well as your posts deleted. This very thing is even covered in the rec.audio.pro FAQ, question 1.9 (marked Q1.8 in the Table of Contents, but it's Q1.9 in the main body): http://recordist.com/rapfaq/
Reply by ●February 27, 20092009-02-27
there's no need to be rude....never. Nobody forces anyone to read and reply. if you consider a question stupid or think it's a homework, just do not reply... Alex
Reply by ●February 27, 20092009-02-27
Alex_001 wrote:> there's no need to be rude....never. Nobody forces anyone to read and > reply. > if you consider a question stupid or think it's a homework, just do not > reply...GFY
Reply by ●February 27, 20092009-02-27
Alex_001 wrote:> there's no need to be rude....never. Nobody forces anyone to read and > reply. > if you consider a question stupid or think it's a homework, just do not > reply...No need, but some get their jollies that way. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
