Values of z are in the range 0 up to 1.5 or 2 currrently. For typical values of the parameters I find that a Taylor series departures significantly above z=0.5. An alternative expansion, with uniform convergence in this range would be needed. I will explore that. Any suggestion is welcome.

I am looking for a nonlinear fit with a chi^2 function that involves integrals like this one: Int(1/sqrt((1+zz)^(2*(1-beta))*(1+sigma*(1+zz)^lambda)^P+B*(1+zz)^3),zz=0..z) Also, for minimization, (some estimation of) the derivatives of the object function wrt the parameters are needed. So it would be convenient to be able to generate code for the object function and its (analytically calculated) derivatives.

I am looking for a nonlinear fit with a chi^2 function that involves integrals like this one: Int(1/sqrt((1+zz)^(2*(1-beta))*(1+sigma*(1+zz)^lambda)^P+B*(1+zz)^3),zz=0..z) Also, for minimization, (some estimation of) the derivatives of the object function wrt the parameters are needed. So it would be convenient to be able to generate code for the object function and its (analytically calculated) derivatives.

I wonder whether it is possible to monitor the stack usage along a calculation. I think that it might be useful e.g. to identify code problems and to stop calculations before they reach the limit. Alejandro

Yes, these stacklimit figures are from personal (nonshared) linux boxes. ulimit figures for the RH7.1 box with 1GB RAM are:

ulimit -a
core file size (blocks)     1000000
data seg size (kbytes)      unlimited
file size (blocks)          unlimited
max locked memory (kbytes)  unlimited
max memory size (kbytes)    unlimited
open files                  1024
pipe size (512 bytes)       8
stack size (kbytes)         8192
cpu time (seconds)          unlimited
max user processes          32767
virtual memory (kbytes)     unlimited

ulimit -H 
unlimited 

ulimit -S 
unlimited

These limits are independent of the amount of RAM on the system. They can be set to whatever you like. However if they are set too large, a poorly written or malicious program could consume all the resources of the system and effectively halt the machine. The stack limits you describe are in kibibytes so a limit of 2937000 is almost 3 Gigs of memory.

So, if I understand correctly there is an "afordable limit" say, determined the amount of RAM and the size of the swap partition, I guess.

Then for 1GB RAM and 1GB swap, what maximum value for stacklimit would you suggest? It seems that default limits are not fine for GOT calculations above some size. I would set limits as high as possible if that allows the calculations to be done (within a few hours).