External module function

The external single ion module has to provide the components nofcomponentsxnofcomponents matrix (see equation (195)) for every transition which is to be taken into consideration in the calculation. Note, in general it is a quadratic matrix with the same dimension as the vectors and Hxc. Actually it is not necessary to compute the whole Matrix , because only the first eigenvector is needed in the computations. Thus, in order to make calculations easier and provide a unique phase of the eigenvectors, the external single ion module must return not the Matrix but the unnormalized eigenvector , which is given by equation (197): . Note that in contrast to the eigenvector is not normalised (no !) and thus the matrix may be recovered from it.

The format to be used is:

extern "C" int du1calc(int & tn,double & T,Vector & Hxc,Vector & Hext,double * g_J,Vector & MODPAR, char ** sipffilename,ComplexVector & u1,float & delta, ComplexMatrix & est)

The meaning of the symbols is as follows:

on input |tn| transition-number sign(tn) >0 standard, <0 routine should do some printout to stdout for user information MODPAR Vector with Parameters read in single ion property file sipffilename file name of the single ion parameter file g_J Lande factor T Temperature[K] Hxc vector of exchange field [meV] (can be n-dimensional, for a set of n operators) Hext external magnetic field [T] est eigenstate matrix (initialized by estates) it should/may also contain population numbers of the states (imaginary part of row 0) and eigenvalues (real part of row 0) with values set by the most recent call for this ion (use of this matrix is optional) u1(1) ninit + i pinit (from mcdisp options -ninit and -pinit) delta upper boundary for transition energy (meV) to be considered (from mcdisp option -maxE) on output int total number of transitions delta transition energy [meV] u1 vector u1=<-|Ialpha-<Ialpha>|+>sqrt((p- - p+)) note that as in Icalc the single ion momentum vector <-|I|+> (n- dimensional with n>=1) may be an arbitrary set of operatorsThe module function must perform the following tasks:

- check if the dimensions of vectors I,Hxc (taken by
*mcphas*from the number of interaction constant columns in*mcphas.j*) and MODPAR (taken by*mcphas*from the number of params in the single ion property file) agree with the module specifications. Note a module may be designed to take different dimensions depending on the input files, however the dimensions of vectors I and Hxc has to agree and must be within the range of dimensions which can be treated by the module. If the check fails the module function should exit the program with an appropriate error message - the module function should do a numbering of all possible single ion transitions and return
the total number of transitions as an integer. Input file parameters params are supplied as a vector MODPAR and
Lande factor as g_J and can be used for this purpose. The numbering will depend on
the parameters ninit, pinit and maxE which are provided as input. These parameters
have to be considered.
*IMPORTANT: the numbering scheme of transitions has to be the same for du1calc and all the corresponding d...1 functions for observables !* - it should calculate from meanfields at a given temperature the
transition energy of transition number
*tn*. The result should be returned as*delta* - for the transition number tn the vector u1 is to be filled with .
- If the energy of this transition is zero, i.e. (diffuse scattering), the expression (195) would be zero because vanishes. In this case the single ion module should calculate instead of .
- if transition is from a level to itself, then a negative value of should be returned.
- if all quantities should be evaluated assuming that all Boltzmann probabilities are zero except for the state number , for which the probability .

Martin Rotter 2017-01-10