Producing incorrect results with abs command

Hey, I am trying to do some pretty basic math here. The equation Im using is which as a i understand would be written f8=abs(x.^x)-2.*x where x is a -10:10 matrix. However, when I run the code, the answers are incorrect. they should start at -20 but the code outputs 0E9. any ideas what I am missing?

[dpb]

Dunno. Attach actual case that you think fails, don't just try to describe it to us...

[Cole]

This is my code output for a a -10:10 matrix.

using a calculator to check those outputs shows it is wrong. for example |-10^-10|-(2*-10) would equal 20.00000000001 not 0 like the output shows. I have to believe it is either the abs function or the subtraction because if I input the values manually, i get the correct answers

Answer 1

[Walter Roberson]

You overlooked the "1.0e+09 *" part of the output. You are using format short, which displays only a limited number of decimal places for each item in order to create a compact display.

x = -10:10;
f8 = abs(x.^x) -2.*x;
f8

f8 = 1×21

1.0e+09 * 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 -0.0000 0 0.0000 0.0000 0.0000 0.0000 0.0008 0.0168 0.3874 10.0000

format long g
f8

f8 = 1×21

1.0e+00 * 20.0000000001 18.0000000025812 16.0000000596046 14.0000012142657 12.0000214334705 10.00032 8.00390625 6.03703703703704 4.25 3 1 -1 0 21 248 3115 46644 823529 16777200 387420471 9999999980

[Cole]

is there a way to translate this to a plot so I dont just show a ton of 0's along the x-axis

[dpb]

With range of 10E9, you can try logarithmic axes and hope for the best...

[Torsten]

@dpb

f8 will be negative in a certain subset of the x-range ...

[dpb]

Well, there's always the FEX submission <symlog>

Answer 2

[dpb]

You didn't provide the input x array so we can (easily) duplicate, but now this one is easy... :)

You're confusing the output display with default format short compared to the actual result stored in memory.

NB: the "1.0e+09 *" at the beginning -- that's a common e-format multiplier applied on all results as printed. The result of the first location thus is reported as "0.000E9" which clearly doesn't have sufficient precision to show a value of 20-something on a 10E9 range.

Try looking at

f8(1)

all by itself and the mystery will go away.