Ensure that random graph generators uses parameter seed

[dcoudert]

We ensure that random graph generators use parameter seed, and we add this parameter if missing.

CC: @kliem @tscrim @vbraun

Component: graph theory

Author: David Coudert

Branch/Commit: 6cbdf91

Reviewer: Travis Scrimshaw

Issue created by migration from https://trac.sagemath.org/ticket/33579

[dcoudert]

Branch: public/graphs/33579_random_seed

[dcoudert]

Commit: a7e1e54

[dcoudert]

comment:1

This might break multiple doctests using random graphs.

---

New commits:

a7e1e54

trac #33579: ensure random graph generators use parameter seed

[dcoudert]

Author: David Coudert

[sagetrac-git]

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

26fe186

trac #33579: fix some doctests

[sagetrac-git]

Changed commit from a7e1e54 to 26fe186

[dcoudert]

comment:3

I'm not sure it's the right way to do it and I don't understand why some doctests have to be modified...

Also, we have some doctest errors in src/sage/tests/books/computational-mathematics-with-sagemath/graphtheory_doctest.py that we will have to fix in the graphtheory.tex (I don't know where it is yet).

**********************************************************************
File "src/sage/tests/books/computational-mathematics-with-sagemath/graphtheory_doctest.py", line 221, in sage.tests.books.computational-mathematics-with-sage\
math.graphtheory_doctest
Failed example:
    max(color.values()) + 1
Expected:
    6
Got:
    5
**********************************************************************
File "src/sage/tests/books/computational-mathematics-with-sagemath/graphtheory_doctest.py", line 226, in sage.tests.books.computational-mathematics-with-sage\
math.graphtheory_doctest
Failed example:
    P = Permutations([0,1,2,3]); P.random_element()
Expected:
    [2, 0, 1, 3]
Got:
    [3, 2, 1, 0]
**********************************************************************
...

[tscrim]

comment:4

My guess as to why doctests are needed to be changed is current_randstate() is not called as often. So all of those doctests that are depending on a specific state are changed. The 32-bit tests will also need to be updated.

For those book tests, you are using a different method of constructing the seed I believe. Hence, it is doing something on a different graph and changing the seed in a different way for the second doctest. So these are essentially trivial failures.

[dcoudert]

comment:5

For those book tests, you are using a different method of constructing the seed I believe. Hence, it is doing something on a different graph and changing the seed in a different way for the second doctest. So these are essentially trivial failures.

I agree these are trivial failures, but I don't know where is to tex file to correct.

[tscrim]

comment:6

I don't know where the tex file is either. Perhaps we can cc (or directly email?) some of the authors to ask?

[sagetrac-git]

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

f0e7684	trac #33579: Merged with 9.6.beta7
2f3f304	trac #33579: fix doctests in sagebook

[sagetrac-git]

Changed commit from 26fe186 to 2f3f304

[dcoudert]

comment:8

I have corrected the doctests in src/sage/tests/books/computational-mathematics-with-sagemath/graphtheory_doctest.py. In parallel I have reported the changes in file graph theory.tex from the sage book (I know have access).

[tscrim]

comment:9

Thanks. Then LGTM.

[tscrim]

Reviewer: Travis Scrimshaw

[dcoudert]

comment:10

One issue remains: I'm unable to check the 32 bits doctests. So we might still have an error.

[tscrim]

comment:11

Volker, if/when this fails on 32-bit, could you please provide what the results should be? We will then fix it right away.

[kliem]

comment:12

Replying to @dcoudert:

I have corrected the doctests in src/sage/tests/books/computational-mathematics-with-sagemath/graphtheory_doctest.py. In parallel I have reported the changes in file graph theory.tex from the sage book (I know have access).

@dcoudert: Could you perhaps apply the same for #32850.

[dcoudert]

comment:13

I will.

[vbraun]

comment:14

I'm getting this on 32-bit:

**********************************************************************
File "src/sage/graphs/generators/random.py", line 99, in sage.graphs.generators.random.RandomGNP
Failed example:
    graphs.RandomGNP(50,.2, algorithm="networkx").size()
Expected:
    260     
Got:
    279
**********************************************************************
1 item had failures:
   1 of  16 in sage.graphs.generators.random.RandomGNP
    [208 tests, 1 failure, 5.27 s]
----------------------------------------------------------------------
sage -t --long --random-seed=0 src/sage/graphs/generators/random.py  # 1 doctest failed
----------------------------------------------------------------------

[sagetrac-git]

Changed commit from 2f3f304 to 6cbdf91

[sagetrac-git]

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

6cbdf91

trac #33579: fix doctest for 32-bit

[dcoudert]

comment:16

Thank you Volker for the value. Should be ok now.

[vbraun]

Changed branch from public/graphs/33579_random_seed to 6cbdf91