This one is trickier than it looks. The other answer doesn't pay it the proper respect.
A general convention is to use the small letter #arccos(x)# and #arcsin(x)# as multivalued expressions, each respectively indicating all the values whose cosine or sine has a given value #x#.
The meaning of the sum of those is really every possible combination, and those wouldn't always give #pi/2.# They won't even always give one of the coterminal angles #\pi/2 + 2pi k quad# integer #k#, as we'll now show.
Let's see how it works with the multivalued inverse trig functions first. Remember in general #\cos x = cos a# has solutions #x=pm a + 2pi k quad# integer #k#.
# c = arccos x# really means
#x = cos c#
#s = arcsin x# really means
#x = sin s#
#y = s + c#
#x# is playing the role of a real parameter that sweeps from #-1# to #1#. We want to solve for #y#, find all the possible values of #y# which have an #x, s# and #c# that makes these simultaneous equations #x = cos c, x = sin s, y = s+c# true.
#sin s = x = cos c #
#cos(\pi/2 - s) = cos c#
We use our above general solution about the equality of cosines.
# pi/2 - s = \pm c + 2pi k quad # integer #k#
# s \pm c = pi/2 - 2pi k #
So we get the much more nebulous result,
#arcsin x pm arcsin c = pi/2 + 2pi k #
(It's permissible to flip the sign on #k.#)
#----------------#
Let's focus now on the principal values, which I write with capital letters:
Show #text{Arc}text{sin}(x) + text{Arc}text{cos}(x) = pi/2#
The statement is indeed true for the principal values defined in the usual way.
The sum is only defined (until we get pretty deep into complex numbers) for #-1 le x le 1# because the valid sines and cosines are in that range.
We'll look at each side of the equivalent
# text{Arc}text{cos}(x) stackrel{?}{=} pi/2 - text{Arc}text{sin}(x)#
We'll take the cosine of both sides.
#cos( text{Arc}text{cos}(x) )= x #
#cos(pi/2 - text{Arc}text{sin}(x)) = sin(text{Arc}text{sin}(x)) =x #
So without worrying about signs or principal values we're sure
#cos( text{Arc}text{cos}(x) )= cos(pi/2 - text{Arc}text{sin}(x)) #
The tricky part, the part that deserves respect, is the next step:
#text{Arc}text{cos}(x) = pi/2 - text{Arc}text{sin}(x) quad# NOT SURE YET
We have to tread carefully. Let's take the positive and negative #x# separately.
First #0 le x le 1#. That means the principal values of both inverse trig functions are in the first quadrant, between #0# and #pi/2.# Constrained to the first quadrant, equal cosines imply equal angles, so we conclude for #x ge 0,#
#text{Arc}text{cos}(x) = pi/2 - text{Arc}text{sin}(x) quad#
Now #-1 le x < 0.# The principal value of the inverse sign is in the fourth quadrant, and for #x < 0# we usually define the principal value in the range
#-\pi/2 le text{Arc}text{sin}(x) < 0#
#\pi/2 ge - text{Arc}text{sin}(x) > 0#
#pi ge pi/2 - text{Arc}text{sin}(x) > pi/2 #
#pi/2 < pi/2 - text{Arc}text{sin}(x) le pi #
The principal value for the negative inverse cosine is the second quadrant,
# \pi/2 < text{Arc}text{cos}(x) le pi#
So we have two angles in the second quadrant whose cosines are equal, and we can conclude the angles are equal. For #x < 0#,
#text{Arc}text{cos}(x) = pi/2 - text{Arc}text{sin}(x) quad#
So either way,
# text{Arc}text{sin}(x) + text{Arc}text{cos}(x) = pi/2 quad sqrt#