Fractional exponents vs. radical form ...

[Albert Chan]

XCas> radpow(a,b) := surd(a^numer(b), denom(b))

numer(b), denom(b) had b simplified first, which is exactly what we wanted
Note: this assumed b is not symbolic variable

XCas> radpow(-64, 2/6)       // = radpow(-64, 1/3) = surd(-64,3) = -4
XCas> radpow(-64, 3/6)       // = radpow(-64, 1/2) = surd(-64,2) = 8*i
XCas> radpow(-64, 4/6)       // = radpow(-64, 2/3) = surd(4096,3) = 16

see https://www.math.ubc.ca/~israel/advisor/.../a37r1.htm