Resistive Möbius Network

Calculus Level 5

Consider a ladder-like network of 2 n 2n nodes (and 3 n 3n branches) in which the extreme nodes are connected with a twist. The given figures, for example, shows the topology for n = 5 n=5 and n = 6 n=6 . Each branch of the network consists of a 1 Ω 1\,\Omega resistor. Let R n R_n denote the effective resistance between any two adjacent side nodes in the network. Evaluate the expression for R n R_n in terms of n n and then find lim n R n \displaystyle\lim_{n\to\infty}R_n .

If the limit is equal to A B ( C + D ) \dfrac{A}{B(C+\sqrt{D})} for coprime positive integers A , B , C , D A,B,C,D , then submit A + B + C + D A+B+C+D .


The answer is 22.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

1 solution

Mark Hennings
Mar 25, 2021

Reorganising the circuit, we consider a loop of 2 n 2n vertices, labelled 1 1 to 2 n 2n , which are connected in a ring. In addition, there are wires connecting vertex j j with vertex n + j n+j for 1 j n 1 \le j \le n . Note that these wires do not meet at the centre of the diagram, but pass by each other without interaction. All wires represent a resistor of size R Ω R\,\Omega . Suppose that a current of I I Amps enters the circuit at vertex 1 1 , and leaves at vertex 2 n 2n . Suppose that the current in the wire from vertex u u to vertex u + 1 u+1 is I u I_u for 1 u 2 n 1 1 \le u \le 2n-1 , and that the current in the wire from vertex 2 n 2n to vertex 1 1 is I 2 n I_{2n} . Suppose also that the current in the wire from vertex u u to vertex n + u n+u is J u J_u for 1 u n 1 \le u \le n . Conservation of current tells us that I 2 n + I = I 1 + J 1 I n + j 1 + J j = I n + j 1 j n 1 I j 1 = I j + J j 2 j n I 2 n 1 + J n = I + I 2 n \begin{array}{rclclcrclcl} I_{2n} + I & = & I_1 + J_1 & & & & I_{n+j-1} + J_j & = & I_{n+j} & \hspace{0.25cm} & 1 \le j \le n-1 \\ I_{j-1} & = & I_j + J_j & \hspace{0.25cm} & 2 \le j \le n & \hspace{0.5cm} & I_{2n-1} + J_n & = &I + I_{2n} \end{array} Thus I j 1 I j = J j = I n + j I n + j 1 I j 1 + I n + j 1 = I j + I n j \begin{aligned} I_{j-1} - I_j & = \; J_j \; = \; I_{n+j} - I_{n+j-1} \\ I_{j-1} + I_{n+j-1} & = \; I_j + I_{n-j} \end{aligned} for all 2 j n 1 2 \le j \le n-1 , so that I j + I n + j = K I_j + I_{n+j} = K for 1 j n 1 1 \le j \le n-1 . Since I 2 n + I I 1 = J 1 = I n + 1 I n I_{2n} + I - I_1 \; =\; J_1 \; = \; I_{n+1} - I_n we deduce that I n + I 2 n = I 1 + I n + 1 I I_n + I_{2n} = I_1 + I_{n+1} - I , and hence I j + I n + j = { K 1 j n 1 K I j = n I_j + I_{n+j} \; = \; \left\{ \begin{array}{lll} K & \hspace{1cm} & 1 \le j \le n-1 \\ K-I & & j=n \end{array} \right.

Suppose now that the voltage at the vertex j j is V j V_j , for 1 j 2 n 1 \le j \le 2n . Then we have V 2 n V 1 = R I 2 n V j V j + 1 = R I j 1 j 2 n 1 V j V n + j = R J j 1 j n \begin{array}{rclll} V_{2n} - V_1 & = \; RI_{2n} \\ V_j - V_{j+1} & = \; RI_j & \hspace{1cm} & 1 \le j \le 2n-1 \\ V_j - V_{n+j} & = \; RJ_j & & 1 \le j \le n \end{array} and so R K = R I j + R I n + j = V j V j + 1 + V n + j V n + j + 1 V j + 1 + V n + j + 1 = V j + V n + j R K \begin{aligned} RK \; = \; RI_j + RI_{n+j} & = \; V_j - V_{j+1} + V_{n+j} - V_{n+j+1} \\ V_{j+1} + V_{n+j+1} & = \; V_j + V_{n+j} - RK \end{aligned} for 1 j n 1 1 \le j\le n-1 , and hence V j + V n + j = A j R K 1 j n V_j + V_{n+j} \; = \; A - jRK \hspace{2cm} 1 \le j \le n for some constant A A . Since R K R I = R I n + R I 2 n = V n V n + 1 + V 2 n V 1 = ( A n R K ) ( A r K ) = ( n 1 ) R K RK - RI \; = \; RI_n + RI_{2n} \; = \; V_n - V_{n+1} + V_{2n} - V_1 \; =\; (A - nRK) - (A - rK) \; = \; -(n-1)RK we deduce that K = I n K= \tfrac{I}{n} , so that (without loss of generality, we can choose A = 0 A=0 ) I j + I n + j = { 1 n I 1 j n 1 n 1 n I j = n V j + V n j = j n I R 1 j n I_j + I_{n+j} \; = \; \left\{ \begin{array}{lll} \frac{1}{n}I & \hspace{0.5cm} & 1 \le j \le n-1 \\ -\tfrac{n-1}{n}I & & j = n \end{array}\right. \hspace{2cm} V_j + V_{n-j} \; = \; -\tfrac{j}{n}IR \hspace{1cm} 1 \le j\le n Solving the equations V j V n + j = R J j V_j - V_{n+j} = RJ_j and V j + V n + j = j n I R V_j + V_{n+j} = -\tfrac{j}{n}IR for 1 j n 1 \le j \le n simultaneously gives V j = 1 2 R ( J j j n I ) V n + j = 1 2 R ( J j + j n I ) 1 j n V_j \; = \; \tfrac12R\left(J_j - \tfrac{j}{n}I\right) \hspace{1cm} V_{n+j} \; = \; -\tfrac12R\left(J_j + \tfrac{j}{n}I\right) \hspace{2cm} 1 \le j \le n Another round of substitution into the given equations yields V j + 1 4 V j + V j 1 = j n I R 2 j n 1 V_{j+1} - 4V_j + V_{j-1} \; = \; -\tfrac{j}{n}IR \hspace{2cm} 2 \le j \le n-1 Solving this recurrence relation, there must be constants P P and Q Q such that V j = P α j + Q α j j 2 n I R V j + n = P α j Q α j j 2 n I R R J j = 2 ( P α j + Q α j ) \begin{aligned} V_j & = \; P\alpha^j + Q\alpha^{-j} - \tfrac{j}{2n}IR \\ V_{j+n} &= \; -P\alpha^j - Q\alpha^{-j} - \tfrac{j}{2n}IR \\ RJ_j & = \; 2\big(P\alpha^j + Q\alpha^{-j}\big) \end{aligned} for 1 j n 1 \le j \le n , where α = 2 + 3 \alpha = 2 + \sqrt{3} . Since R J 1 = I R + V 2 n 2 V 1 + V 2 R J n = V n 1 2 V n + V n + 1 \begin{aligned} RJ_1 & = \; IR + V_{2n} - 2V_1 + V_2 \\ RJ_n & = \; V_{n-1} - 2V_n + V_{n+1} \end{aligned} we can solve a pair of simultaneous equations to obtain P , Q P,Q . If we do this. we obtain that the effective resistance of this circuit is R n = V 1 V 2 n I = 1 6 ( 6 3 + 2 3 1 + ( 2 + 3 ) n 3 n ) R R_n \; =\; \frac{V_1 - V_{2n}}{I} \; =\; \frac16\left(6 - \sqrt{3} + \frac{2\sqrt{3}}{1 + (2+\sqrt{3})^n} - \frac{3}{n}\right)R Putting R = 1 R = 1 , we deduce that lim n R n = 1 6 ( 6 3 ) = 11 2 ( 6 + 3 ) \lim_{n \to \infty} R_n \; =\; \tfrac16(6 - \sqrt{3}) \; = \; \frac{11}{2(6 + \sqrt{3})} making the answer 11 + 2 + 6 + 3 = 22 11 +2 + 6 + 3 = \boxed{22} .

I didn't simplify the circuit and it got very messy. First, this is my sketch:

https://imgur.com/a/jqTYzKL

Current is neither created, nor destroyed. Therefore we get a couple of equations, regarding currents, flowing in different wires. They can be summarized: J k = { I + I 1 I A 1 = I 2 + I B 1 , k = 1 I A k 1 I A k = I B k I B k 1 , 2 k n 1 I 1 I B n 1 = I A n 1 + I 2 I , k = n J_k=\begin{cases} I+I_1-I_{A_1}=I_2+I_{B_1},\ \ \ \ \ \ \ \ \ \ k=1\\ I_{A_{k-1}}-I_{A_k}=I_{B_k}-I_{B_{k-1}},\ \ \ \ \ \ \ \ 2\leq k\leq n-1\\ I_1-I_{B_{n-1}}=I_{A_{n-1}}+I_2-I,\ \ \ \ \ k=n \end{cases} Let V k V_k and φ k \varphi_k denote the potentials of the k t h k^{-th} outer and inner node respectively. Considering the potentials and Ohm's law, we get: J k R = V k φ k J_kR=V_k-\varphi_k I A k R = V k V k + 1 I_{A_k}R=V_k-V_{k+1} I B k R = φ k φ k + 1 I_{B_k}R=\varphi_k-\varphi_{k+1} I 1 R = φ n V 1 I_1R=\varphi_n-V_1 I 2 R = φ 1 V n I_2R=\varphi_1-V_n Now using J k = I A k 1 I A k J_k=I_{A_{k-1}}-I_{A_k} , J k = I B k I B k 1 J_k=I_{B_k}-I_{B_{k-1}} and the above relations, we deduce: φ k + V k 1 3 V k + V k + 1 = 0 \varphi_k+V_{k-1}-3V_k+V_{k+1}=0 V k + φ k 1 3 φ k + φ k + 1 = 0 , V_k+\varphi_{k-1}-3\varphi_k+\varphi_{k+1}=0, Solving this system of recursive equations, I got: V k + 2 6 V k + 1 + 10 V k 6 V k 1 + V k 2 = 0 V_{k+2}-6V_{k+1}+10V_k-6V_{k-1}+V_{k-2}=0 This yields: V k = A α k + B α k + C V_k=A\alpha^k+B\alpha^{-k}+C with α = 2 + 3 \alpha=2+\sqrt{3} . I chose C = 0 C=0 for simplicity, as it doesn't change the potential difference across any two nodes. But substituting back up, I got φ k = V k \varphi_k=-V_k which was surprising (I see that @Mark Hennings got the same result)! The boundary conditions (outer nodes 1 1 and n n ) gave: A = I R 5 α 1 + 5 α n α 2 α 1 n 8 α n 2 8 α 2 n 15 α n 1 + 15 α 1 n α n 3 + α 3 n + 2 α 1 2 α A=IR\frac{5\alpha^{-1}+5\alpha^{-n}-\alpha^{-2}-\alpha^{1-n}}{8\alpha^{n-2}-8\alpha^{2-n}-15\alpha^{n-1}+15\alpha^{1-n}-\alpha^{n-3}+\alpha^{3-n}+2\alpha^{-1}-2\alpha} B = I R 5 α 5 α n + α 2 + α n 1 8 α n 2 8 α 2 n 15 α n 1 + 15 α 1 n α n 3 + α 3 n + 2 α 1 2 α B=IR\frac{-5\alpha-5\alpha^{n}+\alpha^{2}+\alpha^{n-1}}{8\alpha^{n-2}-8\alpha^{2-n}-15\alpha^{n-1}+15\alpha^{1-n}-\alpha^{n-3}+\alpha^{3-n}+2\alpha^{-1}-2\alpha} And finally I got: R n = V 1 V n I = R 10 α 1 2 n 10 α 1 2 α 1 n + 2 α 1 n 2 α 2 2 n + 2 α 2 8 α 2 8 α 2 2 n 15 α 1 + 15 α 1 2 n α 3 + α 3 2 n + 2 α 1 n 2 α 1 n R_n=\frac{V_1-V_n}{I}=R\frac{10\alpha^{1-2n}-10\alpha^{-1}-2\alpha^{-1-n}+2\alpha^{1-n}-2\alpha^{2-2n}+2\alpha^{-2}}{8\alpha^{-2}-8\alpha^{2-2n}-15\alpha^{-1}+15\alpha^{1-2n}-\alpha^{-3}+\alpha^{3-2n}+2\alpha^{-1-n}-2\alpha^{1-n}} As n n\to\infty this simplifies to: R n = R 10 + 2 α 1 15 + 8 α 1 α 2 = 3 1 R_n=R\frac{-10+2\alpha^{-1}}{-15+8\alpha^{-1}-\alpha^{-2}}=\sqrt{3}-1 Any idea what I did wrong will be appreciated. Thanks in advance! Edit: \textbf{Edit:} Could it be that checking with k = 1 k=1 and k = n k=n is incorrect? If so, how can I take into account the boundary conditions?

Veselin Dimov - 2 weeks, 5 days ago

Log in to reply

It looks to me like your first two equations do not account for the Mobius twist. You have an inner n n -gon and an outer n n -gon, and the k k th vertex of the inner ring is connected to the k k vertex of the outer ring.

Mark Hennings - 2 weeks, 5 days ago

Log in to reply

I updated my first comment, have a look at it.

Veselin Dimov - 1 week, 6 days ago

Log in to reply

@Veselin Dimov After substitution k k 2 k\rightarrow k-2 your new recursive relation is of 4'th order, so its general solution should contain four constants. The characteristic polynomial is Q ( z ) = z 4 6 z 3 + 10 z 2 6 z + 1 = ( z 1 ) 2 ( z α ) ( z α 1 ) V k = A α k + B α k + C + k D , 3 k n 2 \begin{aligned} Q(z) &= z^{4} - 6z^3 + 10z^2 - 6z + 1 = (z-1)^{\boxed{2}}(z-\alpha)(z-\alpha^{-1}) &&&\Rightarrow &&&& V_k&=A\alpha^k+B\alpha^{-k} + C\boxed{ + k D}, & 3&\leq k \leq n-2 \end{aligned} As far as I can tell you are missing the boxed linear term. It corresponds to j 2 n I R -\frac{j}{2n}​IR in Mark Henning's solution for V j V_j .


Edit: You may set C = 0 C=0 arbitrarily, but the remaining constants are still defined by the two initial and two final conditions V 1 , V 2 , V n 1 V_1,\:V_2,\:V_{n-1} and V n V_n . One of the four conditions will be redundant as a result of setting C = 0 C=0 , of course, so you end up with a 3x3 system of equations.

I'm not sure how you managed to solve for A , B A,\:B with the missing constant D D - you should have encountered a contradiction somewhere along the way! Maybe you missed the fact your recursive relation is only valid for 3 k n 2 3\leq k\leq n-2 and ignored the conditions for V 2 , V n 1 V_2,\:V_{n-1} ?

Carsten Meyer - 1 week, 6 days ago

Great job actually finding an explicit expression for R n R_n ! I only considered the symmetric case n = 2 N n=2N you can analyse with only half the network to get rid of that Möbius-twist. Repeated Delta-Wye-Transform on one of the half-networks does the rest, but only yields the limit, not the actual expression for R n R_n .

By the way - A , B , C , D A,\:B,\:C,\:D are most definitely not coprime...


Edit: Yes, they actually are coprime (even though not pairwise)... silly mistake of mine!

Carsten Meyer - 2 weeks, 4 days ago

Log in to reply

Yes they are... There is no integer greater than 1 1 that divides them all. On the other hand, they are not pairwise coprime.

Mark Hennings - 2 weeks, 3 days ago

Log in to reply

My silly mistake, I misread the condition as "pairwise coprime". You're right! A classic reminder of "RTFM" - or "Lesen heßt Lösen", the German equivalent ...

Carsten Meyer - 2 weeks, 2 days ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...