please tell me what this words "eps" do in this instruction.i searched it from eps doc but i don't understand the application here.

                                                                        VthL1=-0.25+eps;
VthH1=0.25+eps;

Answer 1

[Azzi Abdelmalek]

In windows command type

eps

ans =

2.2204e-016eps

eps is a predefined variable

[lotus]

it is an integer?with the value of 2.2204^(-16) @ 2.2204e-016

[lotus]

what's the use specifically?is it ok if i don't put it in the instructions above?

[Azzi Abdelmalek]

Maybe it's used to indicate an error. If you need it or no, depends on your code.

[Image Analyst]

Let me try to explain eps better. If your code is trying to compare a floating point number to EXACTLY 0.25 then it may not be able to do that. Why not? See the FAQ: http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F? That's why they put a slight tolerance around it to that it can be anything from (.25 minus a tiny amount) to (0.25 plus a tiny amount). I usually use around 10*eps rather than eps because I don't know if it is even possible to GET a number between 0.25 and (0.25+eps) - it may well be impossible, so that's why I use a multiple of eps.

[lotus]

i got it.thanks

[Walter Roberson]

There are three numbers between 0.25 and 0.25+eps, namely 0.25+eps/4, 0.25+2*eps/4, and 0.25+3*eps/4. These can also be written 0.25+eps(0.25), 0.25+2*eps(0.25), and 0.25+3*eps(0.25).

When you see an expression written as (Constant + eps) then often what should have been written is either (Constant + eps(Constant)) or (Constant * (1+eps)) (though the two expressions are not identical if the Constant <= 0)

Suppose you are using a search routine that allows you to provide a lower bound for a value, and it is built in to the search routine that it will allow values that equal that lower bound, but you only want values greater than the lower bound. The trick is to add eps(value) to the value, thus getting "the first representable number greater than the value" as the lower bound that it will actually use.

Answer 2

[Nikola]

so if do: find x(x==0)=eps, it will find all values of x that are very close to zero? what about "-eps"? what does that minus sign change?

Answer 3

[Nikola]

or it will change all values where x is 0 to very small value?