Indexed variables #179
Open
Indexed variables #179
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 16 commits
September 10, 2018 18:15
Most importantly when it comes to Parameters, is that self.params will still contain all parameters, whereas the attributes indexed_params and unindexed_params will allow access to those types seperately.
…be handled seperately.
…ence method to create indices has been added.
tBuLi
force-pushed
the
indexed_variables
branch
from
134a6b0
to
be64cc1
Compare
An example of this can be a brute-force implementation of least-squares fitting.
pckroon
reviewed
Sep 11, 2018
|
|
||
| # A mapping of the vars to their (perhaps) indexed counterparts | ||
| self.symbol2indexed = {var if isinstance(var, Variable) else var.base: var | ||
| for var in model_dict.keys()} |
There was a problem hiding this comment.
Does this also work with brackets? {(var if ... else var.base): var for var in model_dict.keys()}
|
|
||
| self.dependent_vars = sorted(self.symbol2indexed.keys(), key=sort_func) | ||
| # Ordering is determined by the Variable names, not their indexed | ||
| # version. (e.g. `y`, not `y[i]` to prevent possible ambiguities.). |
There was a problem hiding this comment.
move comment above line. And define sort_func down here.
|
|
||
| # Extract all the params and vars as a sorted, unique list. | ||
| expressions = model_dict.values() | ||
| _params, self.independent_vars = set([]), set([]) | ||
| _params, _independent_vars, _indices = set([]), set([]), set([]) |
There was a problem hiding this comment.
set() instead of set([])?
| vars, params = seperate_symbols(expression) | ||
| # extract symbols from expression, viewing IndexedSymbols as atoms, | ||
| # e.g. returns `y[i]` not `y` when applicable | ||
| indexed_vars, indexed_params = seperate_symbols(expression) |
There was a problem hiding this comment.
add argument for separate_indices
symfit/core/support.py
Outdated
| else: | ||
| vars.append(symbol) | ||
| elif isinstance(symbol, IndexedVariableBase): | ||
| vars.append(symbol) |
There was a problem hiding this comment.
You can clean this if/elif tree by making a dict of {classes: lists}, and sorting based on that. At the end you combine some (possibly also based on lists/sets of classes).
outcomes = {Parameter: [], IndexedParameter: [], IndexedParameterBase: [], ...}
for symbol in func.free_symbols:
for class_, items in outcomes.items():
if isinstance(symbol, class_):
items.append(symbol)etc.
pckroon
reviewed
Sep 11, 2018
| return {expr} | ||
| else: | ||
| # Call this function recursively | ||
| indexed = set([]) |
added 3 commits
September 11, 2018 12:55
It breaks to much sympy compatibility, and writing a function to extract them from an expression turns out to be rather simple once you get the hang of sympy. Tests have also been updated to reflect that IndexedArguments are no longer free_symbols
Additionally, removed Model.chi, Model.chi_jacobian, and its numerical counterparts. This was a remnant from a dark past when we still had to bow to the will of MINPACK. Removed without deprication, as I seriously doubt anybody even knew about this. Symfit itself didn't even use it internally or test for it.
tBuLi
force-pushed
the
indexed_variables
branch
from
aa22073
to
4c7c71f
Compare
|
@pckroon Thanks for the review. Of course, ultimately we should implement those Objectives. However, to give a general solution for least-squares-solving a tensor equation turns out to be non-trivial ;). In order to have an achievable goal, I think this is a good place to draw the line for this PR. |
|
No problem! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
First version of Indexed-variables support! For more info on the possibilities, also see #175.
The goal with this PR is to get to the point where
IndexedVariable's are fully implemented and working, but forIndexedParameter's we just implement the correct infrastructure and handling on Models, but with no means for solving them.This means that after merging this PR, the following things will work:
The following things will be possible for advanced users willing to implement their own subclass of BaseObjective:
{y[i]: A[i, j] * x[j] + b}.y[i] = y0 + a[i] * exp(- b[i] * x[i]).p.s. my apologies for the huge PR