acos acosh adaptsimp asin asinh atan atan2 atanh betafn BFGS cbind cbrt ceil chol_solve chol_solve cholesky copy copy_recycle copysign cos cosh crossprod dbeta dbinom dchisq det dexp df dgamma diag dlnorm dlogis dnbinom dnorm dpois dt dt1 dunif dweibull eigen epsilon erf erfc exp expm1 eye fabs factorial floor fmod for_each_ij_set gammafn gaxpy gradfdif gradfdifls hesscdif hypot intsimp inv inv invpd invpd j0 j1 jacfdif jn kronecker kurtosis kurtosisc ldexp lgamma linesearch1 linesearch2 lnbetafn lndbeta1 lndmvn lndnorm lndt1 lnfactorial lngammafn log log10 log1p logb lu_decomp lu_solve lu_solve matrix_order matrix_style max maxc maxind maxindc mean meanc median medianc min minc minind minindc mode modec nls_broyden ones operator! operator!= operator% operator& operator* operator* operator+ operator- operator- operator/ operator< operator<< operator<= operator== operator> operator>= operator>> operator^ operator| order pbeta pbinom pchisq pexp pf pgamma plnorm plogis pnbinom pnorm pow ppois prod prodc pt punif pweibull qnorm1 qr_decomp qr_solve qr_solve rbind rint row_interchange sd sd sdc selif seqa sgn sin sinh skew skewc sort sortc sqrt sum sumc svd t tan tanh unique var var varc vech xpnd y0 y1 yn zoom

template<matrix_order RO, matrix_style RS, typename T , matrix_order PO, matrix_style PS, typename FUNCTOR , typename RNGTYPE > Matrix<T,RO,RS> scythe::BFGS ( FUNCTOR  fun, const Matrix< T, PO, PS > &  theta, rng< RNGTYPE > &  runif, unsigned int  maxit, T  tolerance, bool  trace = false  ) Find function minimum using the BFGS algorithm. Numerically find the minimum of a function using the BFGS algorithm. Parameters: fun The function to minimize. This function should take one Matrix (vector) argument of type T and return a single value of type T. theta A column vector of parameter values that anchor the 1-dimensional function. runif A random uniform number generator function object (an object that returns a random uniform variate on (0,1) when its () operator is invoked). maxit The maximum number of iterations. tolerance The convergence tolerance. trace Boolean value determining whether BFGS should print to stdout (defaults to false). See also: linesearch1(FUNCTOR fun, const Matrix<T,PO1,PS1>& theta, const Matrix<T,PO2,PS2>& p) linesearch2(FUNCTOR fun, const Matrix<T,PO1,PS1>& theta, const Matrix<T,PO2,PS2>& p, rng<RNGTYPE>& runif) zoom(FUNCTOR fun, T alpha_lo, T alpha_hi, const Matrix<T,PO1,PS1>& theta, const Matrix<T,PO2,PS2>& p) Exceptions: scythe_dimension_error (Level 1) scythe_convergence_error (Level 0) Note: Users will typically wish to implement fun in terms of a functor. Using a functor provides a generic way in which to evaluate functions with more than one parameter. Furthermore, although one can pass a function pointer to this routine, the compiler cannot inline and fully optimize code referenced by function pointers. References gradfdif(), hesscdif(), inv(), scythe::Matrix_base< ORDER, STYLE >::isColVector(), linesearch2(), SCYTHE_CHECK_10, scythe::Matrix_base< ORDER, STYLE >::size(), and t().