"find" function doesn't work

Hi~ I got a weird problem here. See the attached picture. "find" works for some values and doesn't for some others. It could a stupid question but can anybody help? Thanks a lot!

So the wavelength vector "lem" has some value but "find" can't find it.

The following is what happens in the command window. It can find the number value but can't find the variable x which equals that value.

>> x

x =

507.0350

>> jj = find(lem==x)

jj =

   Empty matrix: 1-by-0

>> jj = find(lem==507.0350)

jj =

        3408

Answer 1

[Walter Roberson]

find() works fine. Your array does not contain the value in x. If you were to try

lem(3408) - x

you would find that the result is non-zero.

See this FAQ

[Yuji Zhang]

oh Thanks a lot Walter!

yes you're right.

I got this:

>> lem(3408)-x

ans =

5.6843e-014

Let me carefully check the values.

[Yuji Zhang]

Hi Walter, Hi everybody, could you help more?

I actually don't know how to handle this small difference in value.

>> format('long')

>> x

x =

5.070350000000000e+002

>> lem(3408)

ans =

5.070350000000000e+002

>> lem(3408)-x

ans =

5.684341886080802e-014

[Yuji Zhang]

they are double precision variables. Thanks~

>> whos x

Name Size Bytes Class Attributes

x 1x1 8 double

>> whos lem

Name Size Bytes Class Attributes

lem 1x80001 640008 double

[Yuji Zhang]

Hi~ Thanks for the help, Walter.

I read something about this flow point error here http://www.mathworks.com/help/techdoc/matlab_prog/f2-12135.html?#bqxyrhp.

I currently use this to deal:

>> jj = find(floor(10000000000*(lem-x))==0)

jj =

3408

it works. Thanks! Let me know if there are better ways to solve this problem. Thanks!

[Yuji Zhang]

actually

jj = find(abs(10000000000*(lem-x))<1)

is a better way. because the error could be >0 or <0.

if it was -1e-14, floor(-1e-14)=-1

[Walter Roberson]

Please examine the documentation for eps()