Interaction-free measurement schemes with ideal Mach-Zehnder interferometer
s promised to distinguish absorptive samples with lower average absorption
than simple transmission schemes. We show that this is only true for an ens
emble of two kinds of samples, where one kind is highly absorptive and the
other is highly transmissive. As soon as a third kind of sample with interm
ediate transmission is introduced, but no phase shift is permitted, the cos
t of information gain in terms of absorbed particles in the samples is high
er in the interferometric scheme. We also investigate the general case of s
amples with a continuous range of transmission and phase shift values, such
that an interferometer's ability to measure both sample characteristics ca
n be exploited. With an interferometer the number of principally distinguis
hable samples increases linearly with the number of probe particles, but wi
th a simple transmission setup it increases as the square root. When wishin
g to distinguish twice as many samples from a continuous sample distributio
n with an interferometric scheme, the number of absorbed particles per samp
le only doubles, but it quadruples with a simple transmission scheme.