Raise Exception if DynamicalSystem initialized with coeffs not in given domain

[Joey Lupo]

Currently, code such as

P.<x,y>=ProjectiveSpace(QQ,1) 
f=DynamicalSystem([CC.0*x^2+2*y^2,1*y^2], domain=P)

does not raise an error as it should. We address this by raising a TypeError? if the given polynomials have a different base ring than the domain.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​0834abb	25513:check if endomorphism is already dynamical system in as_dynamical_system
​ebb392f	25523: Add check in DynamicalSystems when a domain is specified that given polynomials have same base ring

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​3c7aae1	25523: Add check in DynamicalSystems when a domain is specified that given polynomials have same base ring
​246c570	25523:base_ring_check_ds
​f7ad143	25523: Modified the check that polynomial shares base ring with given domain by adding coercion

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​28d72ac

25523: Refactored base ring coercion to take into account checks already in method and removed errant apostrophe

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​77bd02d

25523: reset undesirable change in minimal_model code that was necessary for the solution pre-refactoring

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​02ae5ac

25523: Use coord_ring to get PolyRing

[Ben Hutz]

A couple minor things first

• extra blank lines added in affine_ds are not needed

• is_Quotient import is not used

Now for some still weird functionality

• The following gets a typeerror

  P.<x,y>=ProjectiveSpace(ZZ,1)
  f=DynamicalSystem([3*x^2, 1/5*y^2], domain=P)
  

but this gives back a dynamical system over integer ring

P.<x,y>=ProjectiveSpace(ZZ,1)
f=DynamicalSystem([3*x^2, 1/5*y^2])

• when you pass in an endomorphism the domain isn't used. This may be ok if it is documented this way or maybe we should try to coerce to domain.

  P.<x,y>=ProjectiveSpace(QQ,1)
  P2.<x,y>=ProjectiveSpace(CC,1)
  H = End(P2)
  f=H([CC.0*x^2,y^2])
  g=DynamicalSystem(f, domain=P)
  

• When you don't give it a domain, you can get some very weird behavior

  P.<x,y>=ProjectiveSpace(ZZ,1)
  f=DynamicalSystem([GF(5)(3)*x^2, CC.0*y^2])
  

notice the two coordinate polynomials don't even have the same parent!

f[0].parent(),f[1].parent()

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​07d25b2

25523: generalized coefficient coercion for ds

[Ben Hutz]

A couple issues here.

In projective.ds

R.<x>=QQ[]
DynamicalSystem_projective([x, x+1])

This is a weird error. I think you should do a little sanity checking before creating the domain.

In this one, it is functioning correctly, but I think you should add what domain you are attempting to use to the error message since it is not passed in by the user

R.<x,y,z>=QQ[]
f=DynamicalSystem_projective([x+y+z,y*z])

you need to add a couple doc tests showing the behavior you are correcting, especially an example, where the domain is created based off of the parent of the first coordinate function

---

in affine.ds

definitely missing some checks somewhere in here

R.<x>=QQ[]
DynamicalSystem_affine([x, x+1,2*x])

and

R.<x,y,z>=QQ[]
f=DynamicalSystem_affine([x+y+z,y*z])

notice the dimension of the affine space these are creating as a domain. Need to sanity check the number of polys matches the dimension

affine also needs some doc tests showing the behavior.

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​9c85d31

25523: Check dimension of space and # of polys match

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​bd22bde

25523: reworked domain construction

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​fb6b726

25523: remove unused imports

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​11b5505

25523: misplaced import

[Ben Hutz]

everything seems to be working well here. I made some minor doc updates, added a couple examples, and made one code change. When the domain is not given, you already have the common parent, so it was just a minor optimization.

If you're fine with these changes, we'll go ahead and mark it positive.

---

New commits:

​424a4f9

25523: minor updates and tests added

[Joey Lupo]

This all looks good to me- good to go.