@Thomas Richard thx for your input.

i removed the ini's from Maple18/win.bin, lib and Users.

i changed my text file to this:

interface(prompt=""):
e:=exp(1);
Digits:=15;
print("maple init loaded...");

And placed in  C:\Users\<my_account_mame>

Rebooted Maple and no result.

tried putting it into Maple18/win.bin, lib and Users (one at a time) -> same same

@tomleslie thx Tom.

Sorry i was not more succinct. I meant DirectSearch package.

I did state i was using M18.

You gave me the clue, I did a search on the MP website for " hdb" and saw what i'd known i'd seen b4, 

http://www.mapleprimes.com/posts/200289-Convert-Help-Of-Older-Maple-Version-To-18

@Carl Love it's a pity Maple doesn't have a Mathematica-esq "scan for issues" command.

@tomleslie and Kitonum. so logical.

Unfrortunately, Tom , when I run your code I get "Error, invalid input, expects 2 arguments, but received 1" .

in Maple 18....

@Carl Love thanks

@bfathi .sav are file extensions used by SPSS aern't they?

@Axel Vogt 

the orig expression consists of two gaussians and a decay function:

T1:=(1/2)*exp(-(1/2)*(X[4]-933)^2/63^2)*exp(-(1/2)*(X[2]-805)^2/55^2)/((1+exp((1/50)*X[4]-(1/50)*X[2]))*Pi*63*55);

On the subject of maximums, fsolve({diff(ln(T1), X[2]) = 0, diff(ln(T1), X[4]) = 0}); yields 

evalf(eval(T1, {X[2] = 845.348137692692, X[4]= 880.060572171785]}));

0.00000821.....

BUT best results with:

evalf(eval(T1, {X[2] = 843.611792749488, X[4] = 882.338775066865}));

0.00000819.....

I dont really follow your "fsolve locates it in..x=844.09".methodology. c(x) is one gaussian, f(x) is what? 1/(1+exp(x/50))?

@Markiyan Hirnyk 

@acer don't know why the fsolve didnt work.. im using an older version than you...v18.00, win7, 64 bit

@acer I dont get a numerical answer to your final two lines of code. just returns the above.

using maple 18.

@Carl Love 

the quote comes from pp 25 of  http://www.nessis.org/nessis11/marcus.pdf 

which references pp8-10 http://www.ratingscentral.com/Doc/NewTTRS.pdf 

see also

http://www.ratingscentral.com/HowItWorks.php

BTW I did contact the author......

@Carl Love Fantastic! Here is the result of 3 iterations.

I wonder if you or someone else can assist in speeding up the calculation. according to the paper  "Because integrands are functions of playing-strength difference, integrals are convolutions and may be rapidly calculated via the FFT". any suggestions for my maple sheet?

@Kitonum 

found this at:

http://us4.campaign-archive1.com/?u=6ae8427d1f6f3b8d6da57d629&id=b8168c8f89&e=77fe5c923a

Thanks @Kitonum and C.Love. Happy xmas /new year

@Kitonum 

i'll have to take your word for it.  for 1. when i run it, it takes a long time and i get error "maple cannot allocate enough memory...". i'm using M18, 64bit, 2GHz,8GB ram