<div dir="ltr"><br><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="">
<br>
</div>Typically, you need to tell mpmath to use an appropriate precision for<br>
the evaluation:<br>
<br>
>>> <a href="http://bl-1.com/click/load/AzZebQZnUGVfPQNgU2Q-b0231">mpmath.mp.dps</a> = 300<br>
>>> float(mpmath.hyp2f1('10','5','-300.5','0.5'))<br>
-3.8520270815239185e+32<br>
<div class=""><br>
</div></blockquote><div>Oh k! I didn't pay that much heed to it and set <a href="http://bl-1.com/click/load/UGVbaAdmUmdXNVU2U2c-b0231">mp.dps</a> to 100. Wonder why the<br></div><div>difference was so drastic. Anyways thanks a lot. <br>
</div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><br>
    ***<br>
<br>
What can help in hyp2f1 for large values of a,b,c is use of recurrence<br>
relations. These are typically stable in one direction only. [1] This<br>
seems to be still a partially open research question...<br>
<br>
Our current hyp2f1 implementation does use recurrences (hyp2f1ra), but<br>
perhaps they are not invoked for this case. The problem here can be the<br>
accurate determination of the convergence region for each parameter value.<br>
<br>
[1] <a href="http://bl-1.com/click/load/BzIAM1c2UmcCYFU2U2Y-b0231" target="_blank">http://www.ams.org/journals/mcom/2007-76-259/S0025-5718-07-01918-7/</a><br>
<div class="HOEnZb"><div class="h5"><br>
--<br>
Pauli Virtanen<br>
<br>
<br>
_______________________________________________<br>
NumPy-Discussion mailing list<br>
<a href="mailto:NumPy-Discussion@scipy.org">NumPy-Discussion@scipy.org</a><br>
<a href="http://bl-1.com/click/load/XWgMP1Y3BTAEZlQ3BDY-b0231" target="_blank">http://mail.scipy.org/mailman/listinfo/numpy-discussion</a><br>
</div></div></blockquote></div><br></div><img alt="" src="http://bl-1.com/img/VWUANlQxBjpRNlUxVWs-b0231.gif" height="1" width="1"></div>