2
$\begingroup$

How do I define the following recurrence relation in Mathematica?

$$ q_0 = \sqrt{w + 2 \epsilon J},\ q_n = \frac{\epsilon\left(J_n q_0 + \sum\limits_{m=1}^{n-1} \left(J_{n-m} + J_{n+m}\right)q_m\right)}{\epsilon\left(2J - J_{2n}\right) +w}, n \geq 1 $$

where $J_n := |n|^{1+\alpha}$ and $J = 1$ and $\epsilon = 1$ and $w = 1$.


Attempt:

J[n_, m_, \[Alpha]_] := 1/(n - m)^(1 + \[Alpha]); 

q[n_, \[Epsilon]_, \[Alpha]_, 
  w_] := \[Epsilon] (J[n, 0, \[Alpha]]*q[0] + 
    Sum[(J[n, m, \[Alpha]] + J[n, -m, \[Alpha]])*q[m], {m, 1, 
      n - 1}])/(\[Epsilon] (2*1 - J[2*n, 0, \[Alpha]]) + w)

q[0, 1, 1, 1] = Sqrt[w + 2 \[Epsilon]];

However, this syntax is not correct. For example,

q[5, 1, 1, 1]

yields $\frac{100}{299} \left(\frac{13 q(1)}{144}+\frac{58 q(2)}{441}+\frac{17 q(3)}{64}+\frac{82 q(4)}{81}+\frac{1}{25} \sqrt{w+2 \epsilon }\right)$. This should be a number, but instead the symbolic representation is given. Why are the $q(i)$ not numbers?


Update: New relation. Mimicking the method in the Answer provided below, I obtain a sequence I do not expect.

$$ g_0 = g_1= \sqrt{w+\epsilon(2J - J_1)},\ g_n = \frac{\epsilon\sum\limits_{m=1}^{n-1} \left(J_{n-m} + J_{n+m-1}\right)g_m}{\epsilon\left(2J - J_{2n-1}\right) + w}, n \geq 2$$

LatticeEndIndex = 10;
g[0, \[Epsilon]_, \[Alpha]_, w_] := 
 Sqrt[w + \[Epsilon] (2 J[0, \[Alpha]] - J[1, \[Alpha]])]
g[1, \[Epsilon]_, \[Alpha]_, w_] := 
 Sqrt[w + \[Epsilon] (2 J[0, \[Alpha]] - J[1, \[Alpha]])]
g[n_Integer?Positive, \[Epsilon]_, \[Alpha]_, w_] := 
 g[n, \[Epsilon], \[Alpha], 
   w] = \[Epsilon]  Sum[(J[n - m, \[Alpha]] + J[n + m - 1, \[Alpha]])*
       g[m, \[Epsilon], \[Alpha], w], {m, 1, 
       n - 1}]/(\[Epsilon]   (2*J[0, \[Alpha]] - 
          J[2*n - 1, \[Alpha]]) + w) // Simplify
offsiteSequence = g[#, 1, 1, 1] & /@ Range[0, LatticeEndIndex ]

This yields $$\left\{\sqrt{2},0,0,0,0,0,0,0,0,0,0\right\}$$ However, I expected this to be a sequence of non-zero values. In particular, since $g_0 = g_1$, then why is the second slot of the sequence vanishing?

$\endgroup$
2
  • $\begingroup$ In the definition of onsiteSequence you used q where you presumably intended to use g $\endgroup$ Commented Jul 22, 2024 at 20:50
  • $\begingroup$ @BobHanlon Thanks, and the main error I was having was that I had needed to use the Clear command to get rid of some interfering variables. $\endgroup$ Commented Jul 22, 2024 at 21:10

1 Answer 1

6
$\begingroup$
$Version

(* "14.0.0 for Mac OS X ARM (64-bit) (December 13, 2023)" *)

Clear["Global`*"]

With memoization

J[0, α_] = 1;

J[n_Integer?Positive, α_] := J[n, α] = Abs[n]^(1 + α);

q[0, ϵ_, α_, w_] := Sqrt[w + 2 ϵ  J[0, α]];

q[n_Integer?Positive, ϵ_, α_, w_] := q[n, ϵ, α, w] =
  ϵ  (J[n, α]*q[0, ϵ, α, w] + 
       Sum[(J[n - m, α] + J[n + m, α])*
         q[m, ϵ, α, w], {m, 1, n - 1}])/(ϵ  (2*J[0, α] - J[2*n, α]) + w) // 
    Simplify

q[5, 1, 1, 1]

(* (7633 Sqrt[3])/846131 *)

seq = q[#, 1, 1, 1] & /@ Range[0, 6]

(* {Sqrt[3], -Sqrt[3], (6 Sqrt[3])/13, -(1/(11 Sqrt[3])), 452/(
 8723 Sqrt[3]), (7633 Sqrt[3])/846131, 2303090/(119304471 Sqrt[3])} *)

seq // N

(* {1.73205, -1.73205, 0.799408, -0.0524864, 0.0299166, 0.0156249, \
0.0111453} *)
$\endgroup$
0

Your Answer

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

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.