[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Is this list still active?

From: Macy
Subject: Re: Is this list still active?
Date: Wed, 2 Oct 2013 16:14:29 -0700

--- address@hidden wrote:

From: "c." <address@hidden>
To: Macy <address@hidden>
Cc: <address@hidden>
Subject: Re: Is this list still active?
Date: Wed, 2 Oct 2013 19:25:51 +0200

On 2 Oct 2013, at 19:06, Macy <address@hidden> wrote:

> Carlo, 
> Thank you for the URL.
> I received a copy from the group's system, so assumed it also went to 
> everybody.
> Perhaps, too easy a question?

This is the latest message from you I can see on the list:

Is that the question you refer to? 
It seems that message did receive an answer though ...


= = = = =
Am I supposed to be bottom posting, too? 

I don't think that was the last one. However, I did get great suggestions for 
that one that resulted in a 100 to 1 speed increase!

After vectorizing the loop, I then checked it for accuracy it is THAT question 
I heard nothing from.  

The message [which I don't have copy of] compares the for i=1:?? loop's 
accuracy to the vectorized loop's accuracy [I always do these sanity checks] 
Usually when I plot sigA-sigB I get a zero line, but in this case I got a very 
interesting random error plot, obviously caused by computer truncation 
accuracies. I expected some error, but what I got starts small, then 'step' 
increases, then 'step' increases again, and so on.  The original error is on 
the order of 1e-7 and the 'error on the error' is around 1e-7, which means YES 
1e-15 is the truncation accuracy in the machin, but what amazed me was the step 
step step of that error.

I may still have the program - I did a stand alone script of both techniques, 
followed by a plot of the difference of the two techniques, hoping someone 
would try the script and tell me what their machine did.  

I hope it doesn't get kicked out for being long message.  Will look for it.

Essentially it is a comparison between using a for loop versus the vectorized 
approach to find the effect of clock jitter on a signal 
then the two results are subtracted [SHOULD BE ZERO], but are not.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]