feat: Add random state feature. #150

john-halloran · 2025-06-06T19:48:37Z

feat: Added random_state feature for reproducibility.

sbillinge

This is great!

We have to decide how much testing we will add. Ideal is 100% coverage, optimal is probably less.

Maybe write the docstrings so I can understand what the class does, then we can decide what to test?

src/diffpy/snmf/snmf_class.py

sbillinge · 2025-06-06T20:59:08Z

src/diffpy/snmf/snmf_class.py

+        MM,
+        Y0=None,
+        X0=None,
+        A=None,


more descriptive name?

There are many different standards for what to name these matrices. Zero agreement between sources that use NMF. I'm inclined to eventually use what sklearn.decomposition.non_negative_factorization uses, which would mean MM->X, X->W, Y->H. But I'd like to leave this as is for the moment until there's a consensus about what would be the most clear or standard. If people will be finding this tool from the sNMF paper, there's also an argument for using the X, Y, and A names because that was used there.

OK, sounds good. It has to be very good reason to break PEP8. The only good enough reason I can think of is to be consistent with scikit-learn. Another way of saying it is that we can "adopt the scikit-learn standard"

sbillinge · 2025-06-06T21:00:48Z

src/diffpy/snmf/snmf_class.py

@@ -4,8 +4,20 @@


 class SNMFOptimizer:
-    def __init__(self, MM, Y0=None, X0=None, A=None, rho=1e12, eta=610, max_iter=500, tol=5e-7, components=None):


we need a docstring here and in the init. Please see scikit-package FAQ about how to write these. Also, look at Yucong's code or diffpy.utils?

Added one here. The package init dates back to the old codebase, but as soon as that is updated it will get a docstring as well.

The package init (i.e., the __init__.py) doesn't need a docstring.

sbillinge · 2025-06-06T21:01:50Z

src/diffpy/snmf/snmf_class.py

@@ -15,23 +27,22 @@ def __init__(self, MM, Y0=None, X0=None, A=None, rho=1e12, eta=610, max_iter=500
        # Capture matrix dimensions
        self.N, self.M = MM.shape
        self.num_updates = 0
+        self.rng = np.random.default_rng(random_state)


can we have a more descriptive variable name? Is this a range? What is the range?

ping on this one.

sbillinge · 2025-06-06T21:02:32Z

src/diffpy/snmf/snmf_class.py

        if self.A is None:
-            self.A = np.ones((self.K, self.M)) + np.random.randn(self.K, self.M) * 1e-3  # Small perturbation
+            self.A = np.ones((self.K, self.M)) + self.rng.normal(0, 1e-3, size=(self.K, self.M))


K and M are probably good names if the matrix decomposition equation is in hte docstring, so they get defined there.

I think you addressed this with your comment to MM, but as a general rule, please respond to each comment so the reviewer knows you have seen it. It wouldn't work here, but just thumbs up works if you have seen a comment and agree, but it saves time in the long run as I don't have to write this long comment...... :)

john-halloran · 2025-06-06T21:09:28Z

This is great!

We have to decide how much testing we will add. Ideal is 100% coverage, optimal is probably less.

Maybe write the docstrings so I can understand what the class does, then we can decide what to test?

Thanks, will work on resolving these. To be clear, for things like the docstrings would you prefer I make new PRs, get those merged, then rebase this one, or just add to this existing PR?

john-halloran · 2025-06-08T06:23:10Z

For now, I will assume anything given as feedback in this PR could be in scope to include.

sbillinge

This is a great start. I left a couple of comments.

sbillinge · 2025-06-08T17:23:57Z

src/diffpy/snmf/snmf_class.py

@@ -17,6 +17,64 @@ def __init__(
        components=None,
        random_state=None,
    ):
+        """Run sNMF based on an ndarray, parameters, and either a number


This is fantastic! Thanks for this. Please see here for our docstring standards, I am not sure if you looked at it:
https://scikit-package.github.io/scikit-package/programming-guides/billinge-group-standards.html#docstrings

For classes it is a bit tricky because what info do we put in the "Class" docstring and what info do we put in the "constructor" (i.e., the __init__()) docstring. After some googling we came up with the breakout that is shown in the DiffractionObjects class that is shown there. We would be after something similar here.

By way of example, I would probably do like this in this case

def SNMFOptimizer: '''Configuration and methods to run the stretched NMF algorithm, sNMF Instantiating the SNMFOptimizer class runs all the analysis immediately. The results can then be accessed as instance attributes of the class (X, Y, and A). Please see <reference to paper> for more details about the algorithm. Attributes ----------- mm : ndarray The array containing the data to be decomposed. Shape is (length_of_signal, number_of_conditions) y0 : ndarray The array containing initial guesses for the component weights at each stretching condition. Shape is (number_of_components, number_of_conditions ... '''

put future development plans into issues, not in the docstring. Just describe the current behavior. Try and keep it brief but highly informational.

To conform to PEP8 standards I lower-cased the variables. I know they correspond to matrices but we should decide which standard to break. The tie-breaker should probably be scikit-learn. Whatever they do, let's do that. Let's also add a small comment (not in the docstring) to remind ourselves in the future if it breaks PEP8 or it will annoy me every time we revisit it and I will try and change it back......

Conditions on instantiation will go in the constructor docstring.

That one describes the init method so should look more like a function docstring. It would look something like....

def __init__(mm....) '''Initialize a SNMFOptimizer instance and run the optimization Parameters ------------ mm : ndarray The array containing the data to be decomposed. Shape is (length_of_signal, number_of_conditions) y0 : ndarray Optional. Defaults to None The array containing initial guesses for the component weights at each stretching condition. Shape is (number_of_components, number_of_conditions ...

I think there was some text before about how Y0 was required. But if it is required it may be better to make it a required (positional) variable in the constructor and not have it optional. we can discuss design decisions too if you like.

Either Y0 or n_components needs to be provided. Currently, Y0.shape overrides n_components if both are provided, and throws an error if neither are provided. The way scikit-learn is a little more flexible and also allows for an n_components which is different from Y0.shape, although I'm not clear on why you'd want that. But I'm not matching their behavior exactly because the current code doesn't allow that.

scikit-learn actually does break PEP8 to upper-case the matrices

sbillinge

good progress, please see comments.

sbillinge · 2025-06-10T01:24:47Z

src/diffpy/snmf/snmf_class.py

@@ -4,6 +4,18 @@


 class SNMFOptimizer:
+    """A self-contained implementation of the stretched NMF algorithm (sNMF),


This is too long. Needs to be < 80 characters, followed by a blank line.

sbillinge · 2025-06-10T01:25:23Z

src/diffpy/snmf/snmf_class.py

+
+    For more information on sNMF, please reference:
+    Gu, R., Rakita, Y., Lan, L. et al. Stretched non-negative matrix factorization.
+    npj Comput Mater 10, 193 (2024). https://doi.org/10.1038/s41524-024-01377-5


we would normally do a list of Class attributes here. Everything that is self.something. This is obviously strongly overlapped with the arguments of the constructor, as many of the attributes get defined in the constructor, but logically they are different. Here we list and dsecribe the class attributes, there we describe the init function arguments.

sbillinge · 2025-06-10T01:26:27Z

src/diffpy/snmf/snmf_class.py

-        of the class (X, Y, and A). Eventually, this will be changed such
-        that __init__ only prepares for the optimization, which will can then
-        be done using fit_transform.
+        """Initialize an instance of SNMF and run the optimization

        Parameters
        ----------
        MM: ndarray


these need a space before the colon (not sure why we adopted that standard, but we did). So mm : ndarray

sbillinge · 2025-06-10T01:28:49Z

src/diffpy/snmf/snmf_class.py

-            provided.
+            The array containing initial guesses for the component weights
+            at each stretching condition. Shape is (number of components, number of
+            conditions) Must be provided if n_components is not provided. Will override


normally we would raise an exception if two conflicting things are provided (we don't want to guess which is the right one) unless there is a good functional reason to do it another way. We like to avoid "magic" and the current behavior of the code could be "magic". Please raise an exception unless there is a strong reason to do otherwise.

sbillinge · 2025-06-10T01:33:20Z

src/diffpy/snmf/snmf_class.py

-            A stretching factor that influences the decomposition. Zero corresponds to
-            no stretching present. Relatively insensitive and typically adjusted in
-            powers of 10.
+            The float which sets a stretching factor that influences the decomposition.


we don't need to say the type here (float) as it is given above. We can just say "The stretching factor...". The same is actually true above, too. Instead of "the array containing initial guesses", it usually works just as "The initial guesses..."

sbillinge · 2025-06-10T01:39:57Z

src/diffpy/snmf/snmf_class.py

+            The maximum number of times to update each of A, X, and Y before stopping
+            the optimization.
+        tol: float
+            The minimum fractional improvement in the objective function to allow


how about "The convergence threshold. This is the minimum......"

sbillinge · 2025-06-10T01:40:45Z

src/diffpy/snmf/snmf_class.py

+            without terminating the optimization. Note that a minimum of 20 updates
+            are run before this parameter is checked.
+        n_components: int
+            The number of components to attempt to extract from MM. Note that this will


attempt? So sometimes it extracs fewer than n_components when it attempts but doesn't manage?

sbillinge · 2025-06-10T01:42:42Z

src/diffpy/snmf/snmf_class.py

+            be overridden by Y0 if that is provided, but must be provided if no Y0 is
+            provided.
+        random_state: int
+            The integer which acts as a reproducible seed for the initial matrices used in


"The random seed used to initialize". I think the second sentence is useful information, but I think everyone will know what this is. btw, let's cross-check if you didn't already so we are using the names for common variables as scikit-learn.

sbillinge · 2025-06-10T01:43:07Z

src/diffpy/snmf/snmf_class.py

@@ -15,23 +27,22 @@ def __init__(self, MM, Y0=None, X0=None, A=None, rho=1e12, eta=610, max_iter=500
        # Capture matrix dimensions
        self.N, self.M = MM.shape
        self.num_updates = 0
+        self.rng = np.random.default_rng(random_state)


ping on this one.

sbillinge · 2025-06-10T01:45:53Z

src/diffpy/snmf/snmf_class.py

        if self.A is None:
-            self.A = np.ones((self.K, self.M)) + np.random.randn(self.K, self.M) * 1e-3  # Small perturbation
+            self.A = np.ones((self.K, self.M)) + self.rng.normal(0, 1e-3, size=(self.K, self.M))


I think you addressed this with your comment to MM, but as a general rule, please respond to each comment so the reviewer knows you have seen it. It wouldn't work here, but just thumbs up works if you have seen a comment and agree, but it saves time in the long run as I don't have to write this long comment...... :)

feat: Add random state feature.

d8d4e11

sbillinge reviewed Jun 6, 2025

View reviewed changes

Add class docstring

ae45726

components->n_components

3d7c8b6

sbillinge reviewed Jun 8, 2025

View reviewed changes

Updated docstring

a0483b4

sbillinge reviewed Jun 10, 2025

View reviewed changes

		@@ -4,8 +4,20 @@


		class SNMFOptimizer:
		def __init__(self, MM, Y0=None, X0=None, A=None, rho=1e12, eta=610, max_iter=500, tol=5e-7, components=None):

		@@ -4,6 +4,18 @@


		class SNMFOptimizer:
		"""A self-contained implementation of the stretched NMF algorithm (sNMF),

feat: Add random state feature. #150

Are you sure you want to change the base?

feat: Add random state feature. #150

Uh oh!

Conversation

john-halloran commented Jun 6, 2025

Uh oh!

sbillinge left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

john-halloran commented Jun 6, 2025

Uh oh!

john-halloran commented Jun 8, 2025

Uh oh!

sbillinge left a comment

Choose a reason for hiding this comment

Uh oh!

sbillinge Jun 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sbillinge left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

sbillinge Jun 8, 2025 •

edited

Loading