In the course of solving question 134644, I encountered the message
Reduce::ztest1: Unable to decide whether numeric quantity -Log[5381171456512]+Log[(-Sqrt[101] Log2-10 Sqrt[101] Log[5]+101 (Log[<<1>>]+<<1>>+<<1>>)+Sqrt[101] Log[Plus[<<2>>]])^10/((30100251-2278951 Sqrt[101]) (Log2+10 Log[5]-Log[Plus[<<2>>]])^10)] is equal to zero. Assuming it is.
with the numeric quantity in Short form. To obtain the entire expression, I tried
EvaluationData[Reduce[eq3 /. {m -> 0, n -> 0}, {A, c1, d, k}, Reals]]
along the line of an answer by Szabolcs. Indeed, it provided the desired complete expression. But, it also provided about 125 error messages, all apparently internally generated by EvaluationData. Although the desired quantity can be isolated by
EvaluationData[Reduce[eq3 /. {m -> 0, n -> 0}, {A, c1, d, k}, Reals]]
["MessagesExpressions"][[-1, 1, 2]] // ReleaseHold
(* -Log[5381171456512] + Log[(-Sqrt[101] Log[2] - 10 Sqrt[101] Log[5] +
101 (Log[2] + 10 Log[5] - Log[30100251 - 2278951 Sqrt[101]]) +
Sqrt[101] Log[30100251 - 2278951 Sqrt[101]])^10/((30100251 - 2278951 Sqrt[101])
(Log[2] + 10 Log[5] - Log[30100251 - 2278951 Sqrt[101]])^10)] *)
I do not find this work-around particularly satisfying. Two questions:
- Is this undesired
EvaluationDatabehavior a bug? (I believe it is.) - Does an alternative work-around exist, perhaps by instructing
Messagenot to useShorton lengthy expressions.
Edit: Bug apparently fixed in Version 11.1
Reduce no longer produces the warning message given above for Reduce[eq3 /. {m -> 0, n -> 0}, {A, c1, d, k}, Reals], so I could not test
EvaluationData[Reduce[eq3 /. {m -> 0, n -> 0}, {A, c1, d, k}, Reals]]
on the exact same case. I tried a different case, and it worked fine.
Reduceno longer produced the particular warning message mentioned in the question. So, I tried a different case which does produce the warning message, andEvaluationDataworked well. Thanks. – bbgodfrey Mar 19 '17 at 21:33