Problem with simultaneous equations with several variables - population genetics
- To: mathgroup at smc.vnet.net
- Subject: [mg8686] Problem with simultaneous equations with several variables - population genetics
- From: Richard Anderson <richardj.anderson at stonebow.otago.ac.nz>
- Date: Mon, 15 Sep 1997 02:49:12 -0400
- Organization: University of Otago, Dunedin, New Zealand
- Sender: owner-wri-mathgroup at wolfram.com
I am a new Mathematica user. My problem occurs in the context of
modelling genetic systems. I have two equations corresponding to the
change in gene frequencies in females and males respectively, and wish
to search for equilibrium values - i.e. values where the iteration
produces no change in gene frequencis for either males or females.
The Solve command as detailed in the literature works fine for the
simpler models we have attempted:-
Solve [{pfs=pf, pms=pm}, {pm,pf}]
where pfs is the function to give the iterated value for females, pms
for males, and pf and pm are the current gene frequencies in terms of
other parameters.
However as the models become more complex, and more variables are added
to thes functions, we have experienced problems.
It is always the case that two equilibrium points will occur,
corresponding to pf=1, pm=1 and pf=0, pm=0. Only an internal equilibiurm
is of interest to us, should one be present. However, after several
hours of evaluating our more complex models, the Kernel shuts down,
professing lack of memory. We know that in many cases an internal
equilibirum does exist, since we have simulated the systems with C++.
(Even if in a particular case it does not, the two equilibria
corresponding to fixation always exist, and should be returned by
Mathematica). What we seek are precise algebraic definitions of the
internal equilibria.
My question is this : is there any way we can assist Mathematica in
finding such definitions by telling it that we know two of the roots of
whichever cubic it is seeking to solve, so as to aid it in discovering
the third?
Many thanks in advance for any help you may be able to give,
Richard Anderson,
University of Otago.