Matrix powers are annoyingly tricky to keep under control due to the fact
that things to explode or implode rather quickly. In fact the famous quote
from Moler, Van Loan "Unfortunately, the roundoff errors in the mth power
of a matrix, say B^m ,are usually small relative to ||B||^m rather than
||B^
Supposedly can control through env variables but I didn't see any effect
On Wed, Feb 24, 2021, 10:12 AM Charles R Harris
wrote:
>
>
> On Wed, Feb 24, 2021 at 8:02 AM Charles R Harris <
> charlesr.har...@gmail.com> wrote:
>
>>
>>
>> On Wed, Feb 24, 2021 at 5:36 AM Neal Becker wrote:
>>
>>> See m
On Wed, Feb 24, 2021 at 8:02 AM Charles R Harris
wrote:
>
>
> On Wed, Feb 24, 2021 at 5:36 AM Neal Becker wrote:
>
>> See my earlier email - this is fedora 33, python3.9.
>>
>> I'm using fedora 33 standard numpy.
>> ldd says:
>>
>> /usr/lib64/python3.9/site-packages/numpy/core/_
>> multiarray_um
On Wed, Feb 24, 2021 at 5:36 AM Neal Becker wrote:
> See my earlier email - this is fedora 33, python3.9.
>
> I'm using fedora 33 standard numpy.
> ldd says:
>
> /usr/lib64/python3.9/site-packages/numpy/core/_
> multiarray_umath.cpython-39-x86_64-linux-gnu.so:
> linux-vdso.so.1 (0x7ffdd148700
In my experience it is most common to use reasonable but not exceedingly tight bounds in complex applications where there isn’t a proof that the maximum error must be smaller than some number. I would also caution against using a single system to find the tightest tolerance a test passes at. For
See my earlier email - this is fedora 33, python3.9.
I'm using fedora 33 standard numpy.
ldd says:
/usr/lib64/python3.9/site-packages/numpy/core/_multiarray_umath.cpython-39-x86_64-linux-gnu.so:
linux-vdso.so.1 (0x7ffdd1487000)
libflexiblas.so.3 => /lib64/libflexiblas.so.3 (0x7f0512787000
On Wed, Feb 24, 2021 at 11:29 AM Bernard Knaepen wrote:
> Hi all,
>
> We are developing a code that heavily relies on NumPy. Some of our
> regression tests rely on floating point number comparisons and we are a bit
> lost in determining how to choose atol and rtol (we are trying to do all
> opera
Hi all,
We are developing a code that heavily relies on NumPy. Some of our regression
tests rely on floating point number comparisons and we are a bit lost in
determining how to choose atol and rtol (we are trying to do all operations in
double precision). We would like to set atol and rtol as
