Chi^2 calculations above some "size" or "complexity", using Maple 9.5 and Global Optimization Toolbox (GOT), may produce after some time of calculation error messages like:
"Execution stopped: stack limit reached.
The kernel has been shut down. Further computation cannot be performed."
Seeking workarounds, I have looked for information at ?kernelopts for kernelopts(stacklimit), but it was not very useful:
"Limits may be raised or lowered. Maple limits may not be raised above any system defined hard limits. "
Also, a web search produced information only about older versions (I am missing something?).
So, I have issued the commands:
where n is some integer, in several combinations of version, machine (ie RAM), operating system and user interface. The results that I have obtained up to now are collected in the table below.
Here, Cla and Com mean Classic worksheet and Command Line interfaces; default is the result of kernelopts(stacklimit), max is the largest integer not producing an error message of the next column
Maple OS/UI default max error
10 Win/Cla 131072 131072 (1)
10 Win/Com 131072 131072 (1)
10 RH7.1/Cla 8192 2937000 (2)
10 RH7.1/Com 8192 2937000 (2)
9.5 Win/Cla 2048 7907 (1)
9.5 Win/Com 7907 7907 (1)
9.5 RH7.1/Cla 8192 1906501 (3)
9.5 RH7.1/Com 8192 1906501 (3)
9 Win/Cla 2048 7907 (1)
9 Win/Com 7097 7907 (1)
9 RH7.1/Com 8192 1906501 (3)
8 Win/Cla 2048 7907 (4)
8 Win/Com 2016 7907 (4)
(1) Error, cannot raise the stacklimit above the hard limit
(2) Error, too many levels of recursion
(3) Error, integer too large in context
(4) Error, cannot raise the stacklimit
I would be very grateful if you could check these figures in your systems, and fill in missing information to this table.
Provisional conclusions and questions:
These limits seem to be independent of the amount of RAM (in the range 64MB-1GB), so that "hard limit" does not refer to a RAM hardware limit.
Hence, I do not understand from where these different limits arise and whether these four error messages mean the same in "different dialects".
If they were the same, it seems as if that my best chance for avoiding this error is with Maple 10 on Linux, is this true?
Classic and Command interfaces are identical wrt stack limit after Maple 8.
PD I have sent this message to maple-new a few days ago, but there was no reply yet. Perhaps there is a chance in this site. Without any previous message in this forum, I do not know whether this subject is proper here.