Finally, I decided not to depends on SQUAREM, which I used for testing. Nevertherless extrapolation methods for fixed point iteration are at least twice faster than the crude fixed-point iteration or other relaxation methods. Later I would like to do a real benchmark of all methods, not just fixed-point iterations.
Christophe Le 4 déc. 2010 à 15:49, Ravi Varadhan a écrit : > > Hi Christophe, > > Aren't you also using fixed-point acceleration schemes (from the SQUAREM > package) for solving the Nash equilibria? > > How does fixed-point acceleration approach compare to the optimization > approach in terms of speed and robustness (i.e. convergence from bad starting > values)? > > Ravi. > ____________________________________________________________________ > > Ravi Varadhan, Ph.D. > Assistant Professor, > Division of Geriatric Medicine and Gerontology > School of Medicine > Johns Hopkins University > > Ph. (410) 502-2619 > email: [email protected] > > > ----- Original Message ----- > From: Christophe Dutang <[email protected]> > Date: Saturday, December 4, 2010 5:09 am > Subject: Re: [R] "Nash Equilibrium" > To: Ravi Varadhan <[email protected]>, ivo welch <[email protected]> > Cc: r-help <[email protected]> > > >> Hello all, >> >> Months ago, when I wanted to compute Nash equilibria (both standard >> and generalized), I only found one package that implements the >> discrete, i.e. when the player payoff is represented by a matrix. So I >> decided to implement the computation when the payoff is a continuous >> payoff, generally on a compact set. >> >> The project NE computation is available on R-forge (, there is one >> package called GNE computing generalized Nash Equilibria. >> >> Two days ago, I commit a first stable and documented version. Please >> download the binary from the "R packages" tab. The package depends on >> the alabama package, which depends also on numDeriv. So you need to >> download these two packages as well. >> >> Once loaded, ?GNE is an overview of the package. I put 3 examples of >> GNE in the man page taken from von Heusinger & Kanzow (2009). >> >> Christophe >> >> PS: send me an email if you have problems with the functions. >> >> Le 4 déc. 2010 à 00:25, Ravi Varadhan a écrit : >> >>> I think Christophe Dutang is writing a package for generalized Nash >>> Equilibria models called "GNE". >>> >>> I am cc'ing him here. >>> >>> I don't know if there are other packages out there. Christophe >> would know. >>> >>> Ravi. >>> >>> ------------------------------------------------------- >>> Ravi Varadhan, Ph.D. >>> Assistant Professor, >>> Division of Geriatric Medicine and Gerontology School of Medicine Johns >>> Hopkins University >>> >>> Ph. (410) 502-2619 >>> email: [email protected] >>> >>> >>> -----Original Message----- >>> From: [email protected] [ On >>> Behalf Of ivo welch >>> Sent: Friday, December 03, 2010 5:40 PM >>> To: r-help >>> Subject: [R] "Nash Equilibrium" >>> >>> Dear R experts: >>> >>> I searched cran (and r-help) for "nash equilibrium" and "game" but >>> nothing stuck out. has someone written a numerical nash optimizer for >>> two players? >>> >>> player a has choices x1,x2,x3,... and cares about (maximizes) >>> pa(x1,x2,x3,...,y1,y2,y3) >>> player b has choices y1,y2,y3,..., and cares about (maximizes) >>> pb(x1,x2,x3,...,y1,y2,y3) >>> >>> I can tune it to my problem, but if someone has already invented this, >>> please point me to it, so that I do not have to reinvent the wheel. >>> >>> regards, >>> >>> /iaw >>> >>> ---- >>> Ivo Welch ([email protected], [email protected]) >>> >>> ______________________________________________ >>> [email protected] mailing list >>> >>> PLEASE do read the posting guide >>> and provide commented, minimal, self-contained, reproducible code. >>> >> >> -- >> Christophe Dutang >> Ph.D. student at ISFA, Lyon, France >> website: >> >> >> >> >> >> -- Christophe Dutang Ph.D. student at ISFA, Lyon, France website: http://dutangc.free.fr ______________________________________________ [email protected] mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
