Group Abstract Group Abstract

Message Boards Message Boards

0
|
5.2K Views
|
9 Replies
|
1 Total Like
View groups...
Share
Share this post:

[?] Solve problem with Sum function?

Posted 6 years ago

Hi,

I'm trying to find out the value of alpha for which the function w(alpha) is equal to zero. The solve function does not return any value. How can I resolve the issue?

enter image description here

Any assistance will be highly appreciated.

POSTED BY: Matthew Nakeel
9 Replies
In[10]:= Limit[w[\[Alpha]], \[Alpha] -> \[Infinity]]

Out[10]= 0

In[11]:= Limit[w[\[Alpha]], \[Alpha] -> -\[Infinity]]

Out[11]= 0
POSTED BY: Frank Kampas
Posted 6 years ago

Thank you for the insight. Let me try to rephrase the equation so that a single value of [Alpha] can be achieved

POSTED BY: Matthew Nakeel
Posted 6 years ago

Hello Matthew,

for me it is by no means clear what you want to do. What is the w[alpha] you are talking about? In your last post I see only w[ x, y ].

If the "Boundary Conditions" are the equations you want to be fulfilled with a certain alpha that seems to be impossible.

With (giving your w [ x, y ] from above, but as function)

q = 15.904;
v = 0.12157;
E1 = 18904;
b = 1830;
a = 1*b;
h = .04*b;

D1 = (E1 h^3)/(12 (1 - v^2));
w[x_, y_] := 
  q/(24*D1)*(y^4 - 2*b*y^3 + b^3*y) + (2*q)/(D1*b)*
    Evaluate[
     Sum[(((n Pi)/b*x*Sinh[(n Pi)/b*x] - 
           Cosh[(n Pi)/b*
              x]*(2 + (\[Alpha] n Pi)/(2 b)*
               Tanh[(\[Alpha] n Pi)/(2 b)])) Sinh[(n Pi)/b*
           y])/(((n Pi)/b)^5*Cosh[(\[Alpha] n Pi)/(2 b)]), {n, 1, 
       5}]];

some equations are true for any alpha:

w[0, 0]
w[.5 a, 0]

Out[170]= 0

Out[171]= 0

For dealing with other equations I define a new function

f[x_, y_, a_] := w[x, y] /. \[Alpha] -> a

and find

f[-.5 a, b, 1340]
f[-.5 a, b, 1360]

Out[155]= 1.01387*10^6

Out[156]= -459226.

Here we have different signs, therefore I can do a bisection

eps = 10^-6;
xxx = 10;
a1 = 1340;
a2 = 1360;
While[
 xxx > eps,
 a3 = (a1 + a2)/2;
 If[
  Sign[f[.5 a, b, a3]] == Sign[f[.5 a, b, a2]],
  a2 = a3,
  a1 = a3];
 xxx = Abs[a2 - a1]
 ]
a3 // N
f[.5 a, b, a3]

1353.4

0.0386934

0.0386 is not zero, ok, but decreasing eps should give better values. So w[ .5 a , .5 b ] == 0 is almost fulfilled.

But w [ -.5 a, .5 b ] is not

f[-.5 a, .5 b, a3]

Out[26]= -9041.8

In fact I doubt that you can achieve all your "Boundary Conditions" with a single alpha

POSTED BY: Updating Name

boundary conditions are for differential equations

POSTED BY: Frank Kampas

Empty brackets means no solution exists. For example

Solve[x + 1 == x, x]
POSTED BY: Gianluca Gorni
Posted 6 years ago

Apologies. Here's the code.

q = 15.904;
v = 0.12157;
E1 = 18904;
b = 1830;
a = 1*b;
h = .04*b;

D1 = (E1 h^3)/(12 (1 - v^2)); 
w[x_, y_] = 
  q/(24*D1)*(y^4 - 2* b * y^3 + b^3 * y) + (2*q)/(D1*b) * 
    Evaluate[
     Sum[( ((n Pi)/b * x  *Sinh[(n Pi)/b * x] - 
         Cosh[(n Pi)/b * 
            x]*(2 + (\[Alpha] n Pi)/(2 b)* 
             Tanh[(\[Alpha] n Pi)/(2 b)]))  Sinh[(n Pi)/b* y])/(((
        n Pi)/b)^5 * Cosh[(\[Alpha] n Pi)/(2 b)]), {n, 1, 5}]];
(*Boundary conditions*)
w[0, 0] == 0;
w[0, b] == 0;
w[-0.5*a, b] == 0;
w[0.5*a, b] == 0;
w[-0.5*a, 0] == 0;
w[0.5*a, 0] == 0;
w[-0.5*a, 0.5*b] == 0;
w[0.5*a, 0.5*b] == 0;
POSTED BY: Matthew Nakeel

You should use the Code Sample icon (first one) when posting code.

POSTED BY: Frank Kampas
Posted 6 years ago

Hi, thanks for the input. However, the Solve function still returns empty curly brackets { }. When I plot the value of w[\Alpha] against Alpha, the plot is asymptotic to both the x and y axes. This is the entire equation for which I'm trying to find the value of Alphaenter image description here

x varies from -0.5a to 0.5a, while y varies from 0 to b. Is there an error in my syntax or what alternatives can I use?

POSTED BY: Matthew Nakeel

For transcendental equation it is better to give bounds for the variables:

Solve[w[\[Alpha]] == 0 && 0 < \[Alpha] < 100, \[Alpha], Reals]

By the way, the delayed evaluation (:=) cancels out with Evaluate. Immediate evaluation (=) is simpler and (I expect) equivalent.

POSTED BY: Gianluca Gorni
Reply to this discussion
Community posts can be styled and formatted using the Markdown syntax.
Reply Preview
Attachments
Remove
or Discard