[x,y] = ginput(3);
msgbox(sprintf('Value 1 set'));
pause(1.9)
[x2,y2] = ginput(3);
msgbox(sprintf('Value 2 set'));
pause(1.9)
[x3,y3] = ginput(3);
msgbox(sprintf('Value 3 set'));
pause(1.9)
[x4,y4] = ginput(3);
msgbox(sprintf('Value 4 set'));
pause(1.9)
msgbox(sprintf('Plotting...'));
pause(2.5)
msgbox(sprintf('Plotted!'));
plot(x,y,x2,y2,x3,y3,x4,y4)
h = rotate3d;
set(h,'RotateStyle','box','Enable','on');

eps for your system is something like 2.2e-16 most likely, so you are seeing differences because of the details of the numerics. Break the two expressions into pieces and evaluate in chunks (e.g., (1+1/x-1) and 1/(1+1/x-1) vs. the expression in your first statement. I get 1.11022302462516e-015 for (1+1/x-1) instead of what I think you’re expecting 1e-15 because of the order of operations – the difference in between the 2 expressions is about eps/2.

When you are working at this margin, you need to be careful with calculations and the order in which they are done. Numerically, they are not associative.

–Loren

res=3.03..

while when making x=10^13, I get a better evaluation: 2.71..
when mathematically isnt making sense.

In addition, when trying this equivalent expression:
x=10^15
res=(1+1/x)^(1/(1+1/x-1))
I get res = 2.71.. which is way better then the equivalent expression above.
Do you have an explanation for this?

Thanks,
Adam

The bitops originally worked on doubles since that’s all MATLAB had. When the other numeric datatypes were added, the bitops were expanded to the integer types which were the intended input values. Singles got passed by. Perhaps you could create an enhancement request for this functionality, explaining the situation when the need arises for you.

–Loren

It is a bit annoying that bitops on doubles are supoprted out of the box, ex:

>> a=double(1);b=double(2);c=bitor(a,b);c,whos a b c
c = 3
Name Size Bytes Class Attributes
a 1×1 8 double
b 1×1 8 double
c 1×1 8 double

while single-s seem to be less loved:

>> a=single(1);b=single(2);c=bitor(a,b);c,whos a b c
??? Undefined function or method ‘bitor’ for input arguments of type ‘single’.

Of course, one can always

>> a=single(1);b=single(2);c=single(bitor(double(a),double(b)));c,whos a b c
c = 3
Name Size Bytes Class Attributes
a 1×1 4 single
b 1×1 4 single
c 1×1 4 single

but that involves chasing down every single line of user code. :-)

Is there maybe an easy way to “add” bitops support for singles without changing the callee? Like write bitop functions wrappers that accept singles? Expect not but thought I’d better ask just in case.

Thanks in advance for your help.

Regards,
Ljubomir

Perhaps you could take this discussion off-line or with technical support as it is not on-topic for this blog article. Thank you.

-Loren