Equality in Function doesn't work properly

Replies to This Discussion

Permalink Reply by Andrew Heumann on April 28, 2012 at 8:37pm

I'm not 100% sure, but I think the two operate with slightly different tolerances. When I compare 4.0000000000001 and 4, the equality component says they are equal, and the function component says they are not. What values exactly are you giving?

If this is presenting a problem, you might take a look at the "near equality" operator in the function component - As I understand it it only compares up to 10 decimal places. 

Permalink Reply by David Rutten on April 29, 2012 at 3:09am

12 decimal places in fact, so yeah, this is probably a rounding artefact. 

NEVER use equality on floating point values that come from different sources, always compute the difference and compare it to a tolerance you're happy with:

Abs(x-y) < 1e-8

--

David Rutten

david@mcneel.com

Poprad, Slovakia

Permalink Reply by Katerina L on April 29, 2012 at 12:53pm

Thanks!

It seems to work with a tolerance up to  1e-14

However,  I don't see any difference in the input values. They seem to be the same (and by geometrical definition must be the same, because they express the area of the same exactly boundary) and they go until 6 digits

Permalink Reply by Danny Boyes on April 29, 2012 at 1:25pm

The panel only shows is a string representation to 6 dp of that number.

Permalink Reply by Katerina L on April 29, 2012 at 1:44pm

And where can I see the whole number?

Is it a settings matter?

Permalink Reply by Danny Boyes on April 29, 2012 at 2:24pm

there are some numbers that cannot be shown in that sense think root 2 and PI they go on on

you might be able to force the issue with a format function in an fx component.

format("{0:0.00000000000000}",x)

Permalink Reply by David Rutten on April 29, 2012 at 2:40pm

Format("{0:R}", x)

The "R" stands for "Roundtrip", meaning the string representation of the number is guaranteed to be lossless. You may end up with a thoroughly unreadable bit of text though.

Format("{0:0.0###############}", x)

This formatting mask forces a single decimal, and allows for 15 more if they are significant. You can combine any number of "0"s and then "#"s to suit your needs.

--

David Rutten

david@mcneel.com

Poprad, Slovakia

Permalink Reply by Katerina L on April 30, 2012 at 2:10pm

Thanks!