Is there any constant points in exponent function such as e x = x ? Obviously there are no real solutions. But what if we consider complex numbers , are there any constant exponents ?
Find a complex number z such as e z = z and 0 < I m ( z ) < π , 0 < R e ( z ) .
If z = x + i y , where x - real part, y - imaginary part, type x + y as an answer. If you think there is no such complex number, type 0.
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.
There has been a discussion about this problem where a possible approach has been suggested. However the answer was computed instead of being derived analytically. So if you have any ideas how to get a proper answer, feel free to share.
Typically equations involving both x and e x don't have analytical solutions (ie solutions in terms of elementary functions). Wolfram|Alpha gives the answer in terms of the Lambert W function, although this is more or less the definition of that function, so it's a bit of a tautology.
The Newton-Raphson approach used in the discussion clearly works well - I wonder if it leads to any interesting fractal structure?
An alternative computational method is to simply iterate the function. Doing this in the original form blows up rapidly - but note that we also have z = lo g z . We're taking the principal value of the logarithm in order to find a solution in the specified part of the complex plane.
This is most easily iterated by writing z = R e i θ . Then lo g z = lo g R + i θ (this is a naive approach to the complex logarithm but is good enough here). This gives as a map
R → ( lo g R ) 2 + θ 2
and
θ → arctan lo g R θ
Again, a bit of care needs to be taken with the arctan function when implementing this, but with a reasonable initial estimate it converges quickly.
Incidentally, if you replace the above with θ → arctan lo g R θ + 2 k π for integer k , you can converge to different roots outside the range you specified (one example is at around z = 2 . 0 6 2 2 7 7 7 3 + 7 . 5 8 8 6 3 1 1 7 8 i ).
Log in to reply
Greetings. I posted a solution with a graph showing many solutions for z .
Problem Loading...
Note Loading...
Set Loading...
My Newton Raphson solution is posted at the discussion link. I will also include here a plot of many solutions for z , based on a slightly modified version of the posted code. Two things stand out:
1) Solutions consist of complex conjugate pairs
2) The axis scalings for the real and imaginary parts of z are very different
There is a slight asymmetry between the upper and lower halves of the graph, but this apparent asymmetry is almost certainly due to the randomness of the trials I simulated.