Matlab solver for natural log equation

조회 수: 27 (최근 30일)
Shane Palmer
Shane Palmer 2020년 11월 20일
편집: Walter Roberson 2020년 11월 20일
Hello,
I thought this was a simple solver equation, I know I am doing something wrong, because the result should be about S=1429.
This is my code:
syms x
eqn = log(0.5) == (-27109/x)-0.95*log(x)+25.18
S = vpasolve(eqn,x)
Matlab gives me a very large S =672919621689.2986708066438355574
What am I doing wrong here?

채택된 답변

John D'Errico
John D'Errico 2020년 11월 20일
편집: John D'Errico 2020년 11월 20일
Does a solution near(er) to zero exist? PLOT IT!!!!!!
syms x
f = (-27109/x)-0.95*log(x)+25.18 - log(0.5);
fplot(f,[10,1000])
Do you see anything strange? A solution would correspond to a point where that curve crosses the line y == 0. Looking a little further up, we finally see this:
fplot(f,[200,1e4])
yline(0);
So a solution seems to occur near x == 1500.
S = vpasolve(f,x,1500)
S = 
1428.9171626682572235629898202448
However, it you look further out, the curve reaches a peak, and then starts to drop again. It may well cross zero again out there.
fplot(f,[10000,1e6])
yline(0);
Did you tell vpasolve which root it should find? Should it know?
vpasolve is a numerical root finder. It looks for a root, near where you tell it to look. If you don't tell it where, it picks a spot. And sometimes, it may end up converging to a solution you don't like.

추가 답변 (1개)

Walter Roberson
Walter Roberson 2020년 11월 20일
편집: Walter Roberson 2020년 11월 20일
Q = @sym; %convert to rational
syms x
eqn = log(Q(0.5)) == (-Q(27109)/x)-Q(0.95)*log(x)+Q(25.18)
eqn = 
S = solve(eqn,x)
S = 
vpa(S)
ans = 
You can see from this that there are two solutions. You could use an initial value in vpasolve to get the other one

카테고리

Help CenterFile Exchange에서 Numbers and Precision에 대해 자세히 알아보기

제품

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by