Question

An IP Subnet is defined with two parts, a network and a prefix-length or mask.
For example 192.168.0.0/16 (or, 192.168.0.0/255.255.0.0).

An IP address like 192.168.1.1 is said to match this subnet because,

(192.168.1.1 & 255.255.0.0) == 192.168.0.0

I am interested in what might be called the inverse of a subnet
which is described like this,

For a given SubnetA (say, NetworkA / MaskA),
The inverse of SubnetA is the list of k subnets, such that,

If an IP address A, matches SubnetA,
A will not match any of these k subnets, and
Every IP address B that does not match SubnetA,
will match exactly 1 of these k subnets.

Code is not necessary, I am interested in a correct and optimal method.

---

I have the optimized answer noted for reference below so it does not distract people trying this as a problem. Have retained acceptance of Rafał’s answer since he also got it right first.

Answer 1

One subnet for every unmasked bit b in A, matching all the previous bits in A, differing on b, masking all the following bits. This way each address i not in A will match only one of the above networks, namely the one responsible for the first bit of i that does not match A.