search

seacode / gmacs

A generic size-structured stock assessment model
https://seacode.github.io/gmacs
19 stars 14 forks source link

Accounting for new and old shell #30

Closed smartell closed 9 years ago

smartell commented 9 years ago

Need to add accounting for new shell and old shell. Ensure that old shell crabs are not multiplied by the size transition matrix, and the new shell crabs do grow.

smartell commented 9 years ago

Dec 11, 2014. Martell & Ianelli at snowgoose. We had a discussion regarding how to deal with the joint probability of molting and growing to a new size interval for a given length, and the probability of not molting. We settled on using the size-tranistion matrix to represent this joint probability, where the diagonal of the matrix to represent the probability of surviving and molting to a new size interval. The upper diagonal of the size-transition matrix represent the probability of growing to size interval j' given size interval j. Oldshell crabs are then the column vector of 1-molt_probabiltiy times the numbers-at-length, and the Newshell crabs is the column vector of molt_probability times the number-at-length.

I tested this out in a little cpp program, and this seems to work. Its not clear what sort of confounding will result between molting probability parameters and growth parameters, but nonetheless, it greatly simplifies the mathematics and accounting for new shell and old shell crabs. It is also logically consistent.

awhitten commented 9 years ago

Good job snowgeese!

On Fri, 12 Dec 2014 10:22 Steve Martell notifications@github.com wrote:

Dec 11, 2014. Martell & Ianelli at snowgoose. We had a discussion regarding how to deal with the joint probability of molting and growing to a new size interval for a given length, and the probability of not molting. We settled on using the size-tranistion matrix to represent this joint probability, where the diagonal of the matrix to represent the probability of surviving and molting to a new size interval. The upper diagonal of the size-transition matrix represent the probability of growing to size interval j' given size interval j. Oldshell crabs are then the column vector of 1-molt_probabiltiy times the numbers-at-length, and the Newshell crabs is the column vector of molt_probability times the number-at-length.

I tested this out in a little cpp program, and this seems to work. Its not clear what sort of confounding will result between molting probability parameters and growth parameters, but nonetheless, it greatly simplifies the mathematics and accounting for new shell and old shell crabs. It is also logically consistent.

— Reply to this email directly or view it on GitHub https://github.com/seacode/gmacs/issues/30#issuecomment-66710816.

wStockhausen commented 9 years ago

Not sure I understand this. New/old shell should be part of what characterizes the population state as an explicit factor: n(y,x,m,s,z), where x is sex, m is maturity state, s is shell condition, z is size, and y is year in my notation. Then 1-pr(molt|x,m,pre molt z) represents the transition probability of new shell to old shell given sex, maturity state, and size whereas pr(post molt z|pre molt z) represents the size transition for molting new shell to new shell. For KC, not chionoecetes because of terminal molt, you'll have analogous relations for old shell to new shell (molt) and old shell to old shell (no molt). For snow and tanner, old shell is a "terminal" state, of course.

While its possible, I think, to carry out your approach of combining molting prob and growth into a single transition matrix (if I understand what you're talking about) by something similar to what I think you're talking about, this would necessitate combining size and shell condition into a combined factor level.

Of course, given my limited understanding of Gmacs at this point, the preceding remarks may be completely off base (Steve seems to be changing an amazing number of things amazingly quickly).

Buck

On Saturday, December 13, 2014, Steve Martell notifications@github.com wrote:

Closed #30 https://github.com/seacode/gmacs/issues/30.

Reply to this email directly or view it on GitHub https://github.com/seacode/gmacs/issues/30#event-207222220.