3
$\begingroup$

I am attempting a to simulate an 'Enigma M3' machine. As I understand it, the M3 is internally wired as:

Name ABCDEFGHIJKLMNOPQRSTUVWXYZ Notch
I EKMFLGDQVZNTOWYHXUSPAIBRCJ Y
II AJDKSIRUXBLHWTMCQGZNPYFVOE M
III BDFHJLCPRTXVZNYEIWGAKMUSQO D
UKW-B YRUHQSLDPXNGOKMIEBFZCWVJAT

NOTE: I've only included the definitions for the wheels used in this example. As I understand it, the first 3 wheels are the same as the ones used in the Enigma I to enable backwards compatibility.

Given the machine settings:

Name Setting
Wheel Order I II III
Reflector UKW-B
Ring Settings A A A
Wheel Positions A A A

In this configuration, this simulator gives me:

Keyboard Input:       A
Rotors Position:      AAB
Plugboard Encryption: A
Wheel 3 Encryption:   C
Wheel 2 Encryption:   D
Wheel 1 Encryption:   F
Reflector Encryption: S
Wheel 1 Encryption:   S
Wheel 2 Encryption:   E
Wheel 3 Encryption:   B
Plugboard Encryption: B
Output (Lampboard):   B

In my simulator, the following happens:

  • The rightmost wheel (III) advances by one, nets AAB
  • Running A through wheel III at position B nets D instead of C.

The full operation in my simulator:

Operation Result
III D
II K
I N
UKW-B K
I B
II J
III U

What am I doing wrong? Do I have an incorrect definition for wheel III? The above definition of wheel III specifies D in the second space, so D make sense in this context, but appears to be incorrect.

References:

$\endgroup$
1
  • $\begingroup$ I would suggest to get Enigma I working first before venture into M3. Rotors VI, VII and VIII have two turnover positions. They are Z-A and M-N. $\endgroup$ Commented Jan 24 at 17:35

2 Answers 2

2
$\begingroup$

Looks like you've misunderstood what happens to the permutation implemented by a rotor when it rotates. If a rotor which does permutation $\pi$ rotates through one step, the permutation it implements changes to the conjugate $p^{-1}\pi p$ of $\pi$ by the alphabet cycle $p = (A, B, C, \ldots, Z)$.

When rotor III has rotated once, its permutation becomes the conjugate of the one you've given by $p$, so it will send $A \stackrel{p}{\mapsto} B \stackrel{\pi}\to D \stackrel{p^{-1}}\mapsto C$. This is then fed to rotor II which sends C to D, and so on.

You might find this Enigma simulator written in Python, by Brian Neal useful for testing your simulator - generating test cases programatically with that should be a lot easier than running an online simulator "by hand". As far as I know Brian Neal's code is accurate.

$\endgroup$
2
$\begingroup$

Rotors are I, II and III. Rings and indicator settings are at their home positions. Reflector is UKW-B. Suppose $E_0$ is an Enigma I machine without a plugboard; at $t=0$, $n_{\text{indicator}}=1$, i.e. $\texttt{A}$. See footnote. Suppose $E_1$ is the same at $t=1$. We have $E_1 = (\rho^nR\rho^{-n})MLUL^{-1}M^{-1}(\rho^nR^{-1}\rho^{-n})$, where,

$n=n_{\text{indicator}} - n_{\text{ring}} = 1$ (an integer mod 26 in $\mathbb{Z}_{26}$) with $n_{\text{indicator}}=2$, i.e. $\texttt{B}$ and $n_{\text{ring}} = 1$, i.e. $\texttt{01}$;

since $\rho^1 = \rho = \begin{pmatrix} \mathtt{ABCDEFGHIJKLMNOPQRSTUVWXYZ} \\ \mathtt{BCDEFGHIJKLMNOPQRSTUVWXYZA} \end{pmatrix}$,

$\rho^{-1} = \begin{pmatrix} \mathtt{ABCDEFGHIJKLMNOPQRSTUVWXYZ} \\ \mathtt{ZABCDEFGHIJKLMNOPQRSTUVWXY} \end{pmatrix}$ and

$R = \begin{pmatrix} \mathtt{ABCDEFGHIJKLMNOPQRSTUVWXYZ} \\ \mathtt{BDFHJLCPRTXVZNYEIWGAKMUSQO} \end{pmatrix} = \mathtt{(CFLVMZOYQIRWUKXSG)(ABDHPEJT)(N)}$,

we have $\rho R \rho^{-1} = \texttt{(BEKULYNXPHQVTJWRF)(ACGODISZ)(M)}$;

$M = \mathtt{(FIXVYOMW)(CDKLHUP)(ESZ)(BJ)(GR)(NT)(A)(Q)}$;

$L = \mathtt{(AELTPHQXRU)(BKNW)(CMOY)(DFG)(IV)(JZ)(S)}$; and

$U = \mathtt{(AY)(BR)(CU)(DH)(EQ)(FS)(GL)(IP)(JX)(KN)(MO)(TZ)(VW)}$.

Therefore, $E_1 = \texttt{(AB)(CQ)(DM)(EF)(GX)(HI)(JS)(KW)(LP)(NY)(OT)(RV)(UZ)}$.

For example, we can obtain $(\texttt{A})E_1 = \texttt{B}$ from the above expression or via the following diagram.

                B                  A
                ^                  |
                |                  v
rho R rho^-1 = (BEKULYNXPHQVTJWRF)(ACGODISZ)(M)
                 ^                  |
                 |                  |
                 +-----------------+|
                                   ||
                          +--------|+
                          |        |
                          v        |
    M        = (FIXVYOMW)(CDKLHUP)(ESZ)(BJ)(GR)(NT)(A)(Q)
                           |        ^
                           |        |
                           |        +----------------+
                           +------------+            |
                                        |            |
                                        v            |
    L        = (AELTPHQXRU)(BKNW)(CMOY)(DFG)(IV)(JZ)(S)
                                         |           ^
                                    +----+           |
                                    |+---------------+
                                    v|
    U        = (AY)(BR)(CU)(DH)(EQ)(FS)(GL)(IP)(JX)(KN)(MO)(TZ)(VW)

Q.E.D.

Footnote: Indicators $\textbf{n}_{\text{indicator}} = (k_{\text{indicator}}, j_{\text{indicator}}, n_{\text{indicator}})$, rings $\textbf{n}_{\text{ring}} = (k_{\text{ring}}, j_{\text{ring}}, n_{\text{ring}})$, and wiring core offsets = $(k, j, n) = \textbf{n}_{\text{indicator}} - \textbf{n}_{\text{ring}}$ where $k, j, n \in \mathbb{Z}_{26} = \{1, 2, \cdots, 25, 0\}$.

$k_{\text{indicator}}, j_{\text{indicator}}, n_{\text{indicator}} \in \mathbb{A} = \{\mathtt{A}, \mathtt{B}, \cdots, \mathtt{Z}\} \cong \mathbb{Z}_{26}$ and $k_{\text{ring}}, j_{\text{ring}}, n_{\text{ring}} \in \mathbb{B} = \{\mathtt{01}, \mathtt{02}, \cdots, \mathtt{26}\} \cong \mathbb{Z}_{26}$.

P.S. It possibly is not very helpful to keep track of notch positions, which are 8 places ahead of the turnover positions. It is better to use Royal Flags Wave Kings Above or Q-R E-F V-W J-K Z-A.

P.P.S. Expanding the right rotor (III) computation of which you might not be very sure. Turnover position is at V-W. Therefore, there is no middle wheel turnover (mwto) and indicators are $\texttt{AAB}$. $\rho, \rho^{-1}, R$ are elements of a permutation group. For clarity, write $\rho, \rho^{-1}$ in double row notation and $R$ in cycle notation. The latter has three orbits. Their lengths are 17, 8 and 1 respectively.

          AB
          |^
          ||
          v|
rho    = (ABCDEFGHIJKLMNOPQRSTUVWXYZ)
          BCDEFGHIJKLMNOPQRSTUVWXYZA
          |^
          ||
          +|------------------+
           |                  |
          ++                  |
          |                   v
R      = (CFLVMZOYQIRWUKXSG)(ABDHPEJT)(N)
           ^                   |
           |                   |
           +---+               |
               |               |
             +-|---------------+
             | |
             | |
             v |
rho^-1 = (ABCDEFGHIJKLMNOPQRSTUVWXYZ)
          ZABCDEFGHIJKLMNOPQRSTUVWXY
             | ^
             v |
             C E
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.