Never hits the double couple

[vsilwal]

CAP never hits the double couple (ISO=0 and CLVD=0)
The closest we get is ISO 0.000004. It should be okay for inversions but I am not entirely sure how it will effect in the uncertainty estimation. It should not be a big change to fix this.
Thanks,

[celsoa]

the iso parameter parameter has to be interpolated, so it may be close to an expected value but with small differences, in this case 0.000004. there may be a couple of ways to fix this if it turns out to be a problem, but seems like uncertainty estimation needs to be tested first.

It should not be a big change to fix this.

does this refer to the code in the inversion or code in uncertainty estimation?

[vsilwal]

In the inversion code. Even if it does not create any problems in the uncertainty estimation there is no reason we should have this.

[celsoa]

It should not be a big change to fix this.

Sounds like you already have a solution, and sounds like it should not be big. What is your suggestion?

there is no reason we should have this.

We are both working on this project. We both have permissions to make change on the code. Perhaps share your suggestions and we can work together to make the change.

[vsilwal]

If input is a double couple type i.e. -R0/0 and -I1/1/x/y/z, where x,y,z
are number of elements in strike, dip and rake.
In such a case force ISO=0 and CLVD=0 rather than interpolating (this is a
special case). This will require an additional 'if' block.
I am not sure whether that will cause any other search case to break down.
What do you think?

On Fri, Mar 4, 2016 at 4:37 PM, R [email protected] wrote:

It should not be a big change to fix this.

Sounds like you already have a solution, and sounds like it should not be
big. What is your suggestion?

there is no reason we should have this.

We are both working on this project. We both have permissions to make
change on the code. Perhaps share your suggestions and we can work together
to make the change.

—
Reply to this email directly or view it on GitHub
#27 (comment).

Vipul Silwal
Graduate Student
Geophysical Institute
University of Alaska
Fairbanks - 99775, AK
email: [email protected]

[celsoa]

Yes this is the way that CAP worked in the past. But for point solutions there is no reason to go into the inversion block. These can be simply calculated in the main section, I don't think it's a major change or that it would break much. Go for it, or we can discuss next meeting.

[carltape]

A random grid will never hit ISO=0 and CLVD=0 exactly. So this issue is
about the regular grid search. The regular grid is defined by a small
number of points (nv * vw * ...). So loop over the small number of v-w
points and

if abs(v) < TOL, v = 0
if abs(w) < TOL, w = 0

Set TOL to 1e-4 or so.

Perhaps I am missing some intricacies of the implementation. If not, just
do it.

Thanks,
Carl

On Wed, Mar 2, 2016 at 11:27 AM, Vipul Silwal [email protected]
wrote:

CAP never hits the double couple (ISO=0 and CLVD=0)
The closest we get is ISO 0.000004. It should be okay for inversions but I
am not entirely sure how it will effect in the uncertainty estimation. It
should not be a big change to fix this.
Thanks,

—
Reply to this email directly or view it on GitHub
#27.