The description of the ATAN2 function in Microsoft Excel is not 1.

1. The description of the ATAN2 function in Microsoft Excel is not 1.

The description of the ATAN2 function in Microsoft Excel help does not seem
100% accurate to me. It states "ATAN2(a,b) equals ATAN(b/a), except that a
can equal 0 in ATAN2. "

I don't think that is accurate because ATAN2 takes into account the quadrant
in which z lies and has the range of -pi to pi (excluding -pi). This
special inverse tangent uses the sign of x and y to determine the quadrant,
and allows for answers in quandrant 2 (-x, +y) and quadrant 3
(-x,-y) whereas the standard arctan is restricted to quadrants 1 and 4
(range of -pi/2 to +pi/2).

There are cases in which ATAN(y/x) does not equal ATAN2(x,y), other than
when x = 0. For example, when x = -5 and y =-15, ATAN(-15/-5) = 1.249 and
ATAN2(-5,-15) = -1.89

I'm not a math expert, but this description seems misleading to me. I
suggest a description similar to the C++ ATAN2(x,y) function "The atan2()
function computes the arc tangent of y/x, using the signs of the arguments to
compute the quadrant of the return value. "  Register To Reply

2. Re: The description of the ATAN2 function in Microsoft Excel is not 1.

"Celeste" <Celeste@discussions.microsoft.com> wrote...
>The description of the ATAN2 function in Microsoft Excel help does not
>seem 100% accurate to me. It states "ATAN2(a,b) equals ATAN(b/a), except
>that a can equal 0 in ATAN2. "

....

Yes, this is inaccurate except for x and y > 0.

>There are cases in which ATAN(y/x) does not equal ATAN2(x,y), other than
>when x = 0. For example, when x = -5 and y =-15, ATAN(-15/-5) = 1.249 and
>ATAN2(-5,-15) = -1.89

This has everything to do with quadrants. In this case, the ATAN result is
angularly displaced 180 degrees from the ATAN2 result. The ATAN result is
1.249 radians counterclockwise from the positive X axis, and the ATAN2
result is 1.89 radians clockwise from the positive X axis, which happens to
be 1.249 radians counterclockwise from the negative X axis.

>I'm not a math expert, but this description seems misleading to me.
>I suggest a description similar to the C++ ATAN2(x,y) function . . .

You're making the understandable but naive assumption anyone within
Microsoft wants to expend any effort or resources fixing their
documentation. This isn't the only place Excel's documentation (aka online
help) is misleading or just plain wrong.  Register To Reply

3. Re: The description of the ATAN2 function in Microsoft Excel is not 1.

Hi Celeste,

You are right. ATAN2 can do with a better explanation. As can many
other XL functions. {g}

I'll see if someone among my limited contacts within MS -- I am not
particularly astute at networking {vbg} -- is willing to follow up and
improve the documentation. But, I wouldn't hold my breath expecting a
change any time soon.

--
Regards,

Tushar Mehta
www.tushar-mehta.com
Excel, PowerPoint, and VBA add-ins, tutorials
Custom MS Office productivity solutions

In article <35496788-BFED-4210-AC4E-79537480D814@microsoft.com>,
Celeste@discussions.microsoft.com says...
> The description of the ATAN2 function in Microsoft Excel help does not seem
> 100% accurate to me. It states "ATAN2(a,b) equals ATAN(b/a), except that a
> can equal 0 in ATAN2. "
>
> I don't think that is accurate because ATAN2 takes into account the quadrant
> in which z lies and has the range of -pi to pi (excluding -pi). This
> special inverse tangent uses the sign of x and y to determine the quadrant,
> and allows for answers in quandrant 2 (-x, +y) and quadrant 3
> (-x,-y) whereas the standard arctan is restricted to quadrants 1 and 4
> (range of -pi/2 to +pi/2).
>
> There are cases in which ATAN(y/x) does not equal ATAN2(x,y), other than
> when x = 0. For example, when x = -5 and y =-15, ATAN(-15/-5) = 1.249 and
> ATAN2(-5,-15) = -1.89
>
> I'm not a math expert, but this description seems misleading to me. I
> suggest a description similar to the C++ ATAN2(x,y) function "The atan2()
> function computes the arc tangent of y/x, using the signs of the arguments to
> compute the quadrant of the return value. "
>
>  Register To Reply