DSPRelated.com
Forums

LMS maximum step size

Started by cfy30 July 12, 2010

Manny wrote:

> What if you discovered by sheer luck that doing something wanky like > skipping a beat works wonders to your signal? Sure this is uncool, > inelegant, and can't be analyzed, only characterized. Do you scrap all > this and wake up next day pretending that nothing has happened?
It depends. I am cautious about those empirical things. They tend to work great in some cases and fail disgracefully in the other cases. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
On Jul 15, 1:23&#4294967295;am, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
> Manny wrote: > > What if you discovered by sheer luck that doing something wanky like > > skipping a beat works wonders to your signal? Sure this is uncool, > > inelegant, and can't be analyzed, only characterized. Do you scrap all > > this and wake up next day pretending that nothing has happened? > > It depends. I am cautious about those empirical things. They tend to > work great in some cases and fail disgracefully in the other cases.
Noted! Although for completeness---as I now realize I sounded too dramatic above---these ad hoc, empirical techniques are not as blind as they might come across really. I.e. if you know some properties about your expected signal, could rigour be substituted for intuition every blue moon? I don't know the answer for that, and I'm certainly not in a position to advocate for anything here. -Momo
On Jul 14, 7:55&#4294967295;pm, Manny <mlou...@hotmail.com> wrote:
> On Jul 15, 1:23&#4294967295;am, Vladimir Vassilevsky <nos...@nowhere.com> wrote: > > > Manny wrote: > > > What if you discovered by sheer luck that doing something wanky like > > > skipping a beat works wonders to your signal? Sure this is uncool, > > > inelegant, and can't be analyzed, only characterized. Do you scrap all > > > this and wake up next day pretending that nothing has happened? > > > It depends. I am cautious about those empirical things. They tend to > > work great in some cases and fail disgracefully in the other cases. > > Noted! Although for completeness---as I now realize I sounded too > dramatic above---these ad hoc, empirical techniques are not as blind > as they might come across really. I.e. if you know some properties > about your expected signal, could rigour be substituted for intuition > every blue moon? I don't know the answer for that, and I'm certainly > not in a position to advocate for anything here. > > -Momo
Manny, It's certainly not unusual to discover something, and then spend years trying to show why it works or prove why it works. Maurice
On 7/15/2010 8:23 AM, maury wrote:
> On Jul 14, 7:55 pm, Manny<mlou...@hotmail.com> wrote: >> On Jul 15, 1:23 am, Vladimir Vassilevsky<nos...@nowhere.com> wrote: >> >>> Manny wrote: >>>> What if you discovered by sheer luck that doing something wanky like >>>> skipping a beat works wonders to your signal? Sure this is uncool, >>>> inelegant, and can't be analyzed, only characterized. Do you scrap all >>>> this and wake up next day pretending that nothing has happened? >> >>> It depends. I am cautious about those empirical things. They tend to >>> work great in some cases and fail disgracefully in the other cases. >> >> Noted! Although for completeness---as I now realize I sounded too >> dramatic above---these ad hoc, empirical techniques are not as blind >> as they might come across really. I.e. if you know some properties >> about your expected signal, could rigour be substituted for intuition >> every blue moon? I don't know the answer for that, and I'm certainly >> not in a position to advocate for anything here. >> >> -Momo > > Manny, > It's certainly not unusual to discover something, and then spend years > trying to show why it works or prove why it works.
Maury, I'll push that along even further. A lot of junk science comes from people who make up an explanation for something they observe but don't understand.* Sometimes that's even rational. There are people won't believe /what/ unless there's an accompanying /how/. Jerry ______________________________________ * If you screw out a light bulb and put your finger in the hole, it hurts a lot. A spirit lives there and bites you if intrude. -- Engineering is the art of making what you want from things you can get. &#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;&#4294967295;