Comment #7 on issue 3035 by [email protected]: Math.sin(Math.PI)
inaccurate
http://code.google.com/p/v8/issues/detail?id=3035
Every argument made about floating point and accuracy which doesn't involve
the notion of ranges if basically invalid. Every floating point value does
not represent a single mathematical value, but a *range* of values (because
the binary digits beyond the mantisssa are unknown). If you want to argue
about the accuracy of a function, you have to consider how a range of
values is mapped mathematically by the (mathematical) function, and if the
observed result is within that range.
In our case, the first derivative of sine at pi is 1, this means that the
range of values corresponding to Math.PI is mapped to a range of the same
size around the mathematical value of sine(pi) (i.e. 0). As mentioned
above, the IEEE mantissa gives you almost 16 decimal digits of precision,
so anything in the same order around 0 is OK.
--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.
