You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I wrote a testing program, and found that in some cases the error of SZ(v2.1.12.2) exceeds the absolute error bound.
$ ./sz_test < data
data len: 171840
compressed len: 97787
Error at 66559: 116.3225489999999951 vs 116.3225589999999983 (diff: 0.0000100000000032)
The exceeded error is so close to the error bound that I believe it is caused by floating point computation.
I would like to know that whether I'm using SZ in a wrong way, or it is just a bug.
The testing code and data are in the attachments.
Thanks.
I wrote a testing program, and found that in some cases the error of SZ(v2.1.12.2) exceeds the absolute error bound.
The exceeded error is so close to the error bound that I believe it is caused by floating point computation.
I would like to know that whether I'm using SZ in a wrong way, or it is just a bug.
The testing code and data are in the attachments.
Thanks.
CodeAndData.zip
The text was updated successfully, but these errors were encountered: