Question

How can I invert a binary equation, such that I can find which inputs will produce a given output.

Example:

Inputs: i0 through i8
Outputs: o0 through o8
Operators: ^ = XOR, & = AND

Binary equations:

(1&i0) ^ (1&i1) ^ (0&i2) ^ (1&i3) ^ (0&i4) ^ (0&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o0
(0&i0) ^ (1&i1) ^ (0&i2) ^ (1&i3) ^ (1&i4) ^ (0&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o1
(0&i0) ^ (1&i1) ^ (1&i2) ^ (0&i3) ^ (0&i4) ^ (1&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o2
(1&i0) ^ (0&i1) ^ (0&i2) ^ (1&i3) ^ (0&i4) ^ (0&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o3
(0&i0) ^ (1&i1) ^ (0&i2) ^ (1&i3) ^ (1&i4) ^ (0&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o4
(0&i0) ^ (0&i1) ^ (0&i2) ^ (0&i3) ^ (0&i4) ^ (1&i5) ^ (0&i6) ^ (0&i7) ^ (0&i8) = o5
(0&i0) ^ (0&i1) ^ (0&i2) ^ (1&i3) ^ (0&i4) ^ (0&i5) ^ (1&i6) ^ (0&i7) ^ (0&i8) = o6
(0&i0) ^ (0&i1) ^ (0&i2) ^ (1&i3) ^ (1&i4) ^ (0&i5) ^ (1&i6) ^ (1&i7) ^ (0&i8) = o7
(0&i0) ^ (0&i1) ^ (0&i2) ^ (0&i3) ^ (0&i4) ^ (1&i5) ^ (0&i6) ^ (0&i7) ^ (1&i8) = o8

In matrix form:

1,1,0,1,0,0,0,0,0,1
0,1,0,1,1,0,0,0,0,1
0,1,1,0,0,1,0,0,0,1
1,0,0,1,0,0,0,0,0,1
0,1,0,1,1,0,0,0,0,1
0,0,0,0,0,1,0,0,0,1
0,0,0,1,0,0,1,0,0,1
0,0,0,1,1,0,1,1,0,1
0,0,0,0,0,1,0,0,1,1

Additional limitations:

• The prime diagonal is always 1
• I am interested if there is a solution or not, more then the solution itself

How can I, algorithmically find inputs i0 -i8 such that outputs o0 – o8 are 1? What I really want to find is if there is such a solution or not.

I need an algorithm which is scalable to larger networks, at least up to 100 inputs/outputs.

Answer 1

With XOR (rather than OR), it seems at first glance that some form of Gauss–Jordan elimination can at least simplify the problem. Adapting the pseudocode from the Wikipedia article on reduced row echelon form, we get:

function ToReducedRowEchelonForm(Matrix M) is
    // 'lead' is the column index in a row of the leading 1
    lead := 0
    rowCount := the number of rows in M
    columnCount := the number of columns in M
    for 0 ≤ r < rowCount do
        if columnCount ≤ lead then
            stop
        end if
        i = r
        // Find row with lead point
        while M[i, lead] = 0 do
            i = i + 1
            if rowCount = i then
            // no pivot in this column, move to next
                i = r
                lead = lead + 1
                if columnCount = lead then
                    stop
                end if
            end if
        end while
        Swap rows i and r
        for 0 ≤ i < rowCount do
            if i ≠ r do
                Set row i to row i XOR row r
            end if
        end for
        lead = lead + 1
    end for
end function

This converts the sample to:

1,0,0,0,0,0,0,1,0,0
0,1,0,0,0,0,0,0,0,0
0,0,1,0,0,0,0,0,0,0
0,0,0,1,0,0,0,1,0,1
0,0,0,0,1,0,0,1,0,0
0,0,0,0,0,1,0,0,0,1
0,0,0,0,0,0,1,1,0,0
0,0,0,0,0,0,0,0,1,0
0,0,0,0,0,0,0,0,0,0

From there, you could adapt an integer partitioning algorithm to generate the possible inputs for a row, taking in to account the partitions for previous rows. Generating a partition is a great candidate for memoization.

Still need to do a timing analysis to see if the above is worthwhile.