E to the -y plus trial and error math?

  • #1
PhanthomJay
Science Advisor
Homework Helper
Gold Member
7,156
502

Main Question or Discussion Point

A co worker has the following equation:

[itex]3e^{-.5y} +3e^{-y} + 3e^{-1.5y} + 103e^{-2y} =98.39[/itex]

Solve for y.

Some sort of compound interest bond equation I am told, or something like that.

He has been told that to solve for y, one must use a trial and error approach.

True??
 

Answers and Replies

  • #2
1,955
252
If you substitute u for [itex] e^{- \frac {1}{2} y} [/itex] you get a quartic equation, for wich an exact solution exists. Type the equation in WolframAlpha to get a meaningless jumble of really large numbes and lots of square and cube root signs.

It's less work to solve the quartic with a numerical method like newton's method than to use the formula for the quartic equation.
 
  • Like
Likes 1 person
  • #3
D H
Staff Emeritus
Science Advisor
Insights Author
15,393
683
There is an exact solution. Let ##x=\exp(y/2)##. Then your equation is equivalent to ##103x^4+3x^3_3x^2_3x-98.39=0##. This is a quartic equation, so it is solvable, exactly. Then solve for y. Simple!

Not so simple. Solving cubics is a bear of a problem. Solving quartics? That's a megafauna bear of a problem. Solving this numerically is non-trivial. Newton's method doesn't work very well on this problem. You need to use something else such as [strike]the secant method[/strike] the midpoint method.

Edit: The secant method doesn't work very well here either because f(x) is almost flat between -1 and +1.
 
Last edited:
  • #4
PhanthomJay
Science Advisor
Homework Helper
Gold Member
7,156
502
willem2 and DH...thanks! This forum is loaded with some very brilliant minds. He couldn't find an answer anywhere on-line or from his college finance professors, so I told him not to worry, I would get an answer through the best site on the web.

Thanks again!
 
  • #5
D H
Staff Emeritus
Science Advisor
Insights Author
15,393
683
There is an exact solution. Let ##x=\exp(y/2)##. Then your equation is equivalent to ##103x^4+3x^3_3x^2_3x-98.39=0##. This is a quartic equation, so it is solvable, exactly. Then solve for y.
Massive typos there. I should have said

Let ##x=\exp(-y/2)##. Then your equation is equivalent to ##103x^4+3x^3+3x^2+3x-98.39=0##.
 
  • #6
PhanthomJay
Science Advisor
Homework Helper
Gold Member
7,156
502
Massive typos there. I should have said

Let ##x=\exp(-y/2)##. Then your equation is equivalent to ##103x^4+3x^3+3x^2+3x-98.39=0##.
Yes, I realized that afterwards..thanks for the correction!
 
  • #7
9
1
If you substitute u for [itex] e^{- \frac {1}{2} y} [/itex] you get a quartic equation, for wich an exact solution exists. Type the equation in WolframAlpha to get a meaningless jumble of really large numbes and lots of square and cube root signs.

It's less work to solve the quartic with a numerical method like newton's method than to use the formula for the quartic equation.
I attempted this method and recieved reasonable answers of a complex conjugate pair and two real roots. However you obviously cannot ln a negative number which reduces the outcome to three possibilities. Once divided buy the -0.5, ending up with y=0.0677, 0.008-∏i and -13.807+∏i
 
  • Like
Likes 1 person
  • #8
D H
Staff Emeritus
Science Advisor
Insights Author
15,393
683
I attempted this method and recieved reasonable answers of a complex conjugate pair and two real roots. However you obviously cannot ln a negative number which reduces the outcome to three possibilities. Once divided buy the -0.5, ending up with y=0.0677, 0.008-∏i and -13.807+∏i
You can reject the negative solution to ##3(x+x^2+x^3)+103x^4=98.39## (and also the two complex solutions) if you are looking for a real solution to the original equation. You cannot reject that negative solution if you allow complex solutions to the original equation. In fact, each of the four solutions to the polynomial generates an infinite family of complex solutions to the original equation. If some complex valued ##y## is a solution to that original equation, then so is ##y+4k\pi## for any integer k.
 
  • Like
Likes 1 person
  • #9
PhanthomJay
Science Advisor
Homework Helper
Gold Member
7,156
502
The answer given in the text example was in fact y = .0677
 

Related Threads on E to the -y plus trial and error math?

  • Last Post
Replies
2
Views
7K
  • Last Post
Replies
9
Views
3K
  • Last Post
Replies
11
Views
2K
  • Last Post
Replies
2
Views
2K
Replies
1
Views
1K
Replies
1
Views
1K
  • Last Post
Replies
2
Views
3K
  • Last Post
Replies
5
Views
12K
Replies
6
Views
517
Replies
1
Views
780
Top