Completed Parameter Implementation #68

pafloxy · 2023-07-06T11:40:38Z

Before submitting, please check the following:

Make sure you have tests for the new code and that test passes (run pytest)
format added code and tests by black -l 120 <filename>
If applicable, add a line to the [unreleased] part of CHANGELOG.md, following keep-a-changelog.

Then, please fill in below:

Context (if applicable):

Description of the change:
Introduced Parameter class for providing symbolic placeholder for parameterised arguments in circuit (rotation-angles) / patterns (measurement-angles).

It is allowed to make numerical operation on the instances of the Parameter class which upadates the symbolic expression inplace, however interaction between multiple instances of the class are not allowed .

Numerical value can be assigned using Parameter.bind() method

th = Parameter('th');
np.sin(th/3)
------------------
>Out: expr = sin(th[/3])

th.bind({th:0.3})
-------------------
>Out: expr = sin(th[/3]), assignment = [ th : 0.3 ]

th.value
------------------
>Out: 0.09983341664682814

Related issue:
#45

also see that checks (github actions) pass.
If lint check keeps failing, try installing black==22.8.0 as behavior seems to vary across versions.

…ssue45

mgarnier59 · 2023-07-07T08:35:54Z

Do we validate the behaviour modification of output_nodesin pattern.pyaround line 100? I don't yet understand how it works if the output nodes are specified by the user but if it is handled by the pattern, then I fear this modification will break the code since output_nodes is built iteratively.

shinich1 · 2023-07-07T09:30:36Z

Do we validate the behaviour modification of output_nodesin pattern.pyaround line 100? I don't yet understand how it works if the output nodes are specified by the user but if it is handled by the pattern, then I fear this modification will break the code since output_nodes is built iteratively.

Thanks for pointing this out. I agreed to this change in the discussion on the issues page, but I suppose you are right, breaking change like this should be avoided. More so because we don’t need this change for the parametrised patterns to work.

mgarnier59 · 2023-07-07T11:35:38Z

Do we validate the behaviour modification of output_nodesin pattern.pyaround line 100? I don't yet understand how it works if the output nodes are specified by the user but if it is handled by the pattern, then I fear this modification will break the code since output_nodes is built iteratively.

Thanks for pointing this out. I agreed to this change in the discussion on the issues page, but I suppose you are right, breaking change like this should be avoided. More so because we don’t need this change for the parametrised patterns to work.

Maybe the error pointed out by @pafloxy could be resolved by calling the add method directly in __init__.

this may need to be reviewed again with new version of code

shinich1 · 2024-05-08T14:35:51Z

@pafloxy I accidentally deleted the target branch of this PR, sorry!
what should we do with this feature? I think many of the changes I asked back when we were working on this are now likely unnecessary (sorry..) and your implementation may fit very nicely in the current version of graphix? if you feel like it, please open a PR to merge into master.

pafloxy · 2024-05-08T16:23:45Z

Hello @shinich1 ,
Its alright, I think it would be more convenient to rewrite it with the current version of graphix which would also be more convenient. Also we were reconsidering how we use the parameterised class for yourselves among oursleves so maybe this time we will write it in a different fashion, the Parameter class on overall might not be changed but the way it interacts with rest of graphix should be require some re-thinking based on what we decide :)

benjvmin93 · 2024-05-28T11:43:25Z

graphix/parameter.py

+            if overwrite:
+                self._value = float(val)
+                self._assignment = value
+            else:
+                if not self.assigned:
+                    self._value = float(val)
+                    self._assignment = value


In my sense, this part is redundant and should be replaced by something more concise. In any cases, you are setting self._value and self._assignment. For me the conditions should only be if not self.assigned: ... else: raise ValueError("...")

benjvmin93 · 2024-05-28T11:44:02Z

graphix/parameter.py

+                else:
+                    raise ValueError("Symbols are already assigned, set overwrite = True to overwrite value")
+
+        if len(val.free_symbols) != 0:


This could be replaced by an else statement.

benjvmin93 · 2024-05-28T13:17:56Z

graphix/pattern.py

+        if inplace:
+            cmd_indx = -1
+            for cmd in self.seq:
+                cmd_indx += 1
+                if cmd[0] == "M" and isinstance(cmd[3], Parameter):
+                    cmd[3] = cmd[3].bind(parameter_assignment, allow_unknown_parameters=True).value
+                    self.seq[cmd_indx] = cmd
+
+            self.inspect_parameterized_commands()
+
+        else:
+            new_pattern = deepcopy(self)
+            cmd_indx = -1
+            for cmd in new_pattern.seq:
+                cmd_indx += 1
+                if cmd[0] == "M" and isinstance(cmd[3], Parameter):
+                    cmd[3] = cmd[3].bind(parameter_assignment, allow_unknown_parameters=True).value
+                    new_pattern.seq[cmd_indx] = cmd


Here, there is some code duplication which could be avoided.

benjvmin93 · 2024-05-28T13:27:11Z

graphix/transpiler.py

+        if inplace:
+
+            for gate_indx in parameterized_gate_indx:
+
+                gate_instr = self.instruction[gate_indx]
+                if isinstance(gate_instr[-1], (Parameter)):
+                    gate_instr[-1] = gate_instr[-1].bind(parameter_assignment, allow_unknown_parameters=True).value
+                    print(gate_instr)  ##checkflag
+                    self.instruction[gate_indx] = gate_instr
+
+            self.inspect_parameterized_gates()
+
+        else:
+
+            new_pattern = deepcopy(self)
+
+            for gate_indx in parameterized_gate_indx:
+
+                gate_instr = new_pattern.instruction[gate_indx]
+                if isinstance(gate_instr[-1], (Parameter)):
+                    gate_instr[-1] = gate_instr[-1].bind(parameter_assignment, allow_unknown_parameters=True).value
+                    new_pattern.instruction[gate_indx] = gate_instr
+
+            new_pattern.inspect_parameterized_gates()


Code duplication that could be avoided by just assigning self.instruction[gate_indx] or new_pattern.instruction[gate_indx] to gate_instr depending on the inplace condition.

benjvmin93 · 2024-05-28T13:44:10Z

graphix/transpiler.py

+        if self.num_parameterized_gates == 0:
+            raise ValueError(" no paramaters to assign ")
+
+        elif self.num_parameterized_gates > 0:


No need to make this part as a condition statement since we know the num_parameterized_gates attribute is setted to 0 and we already test it in the if statement above and that it leads to an error raising if so.

benjvmin93 · 2024-05-28T13:49:34Z

graphix/transpiler.py

        """
        assert qubit in np.arange(self.width)
+        angle = 1.0 * angle


I don't understand here and in the other rotation instructions why it is necessary to multiply the angle by 1.0. Wouldn't it be interpreted as a parameter object instead of a float anyway? Could you please explain a little more on these lines?

This commit is yet another tentative to implement parameterized measurement patterns, to fulfill issue TeamGraphix#45 (previous tentative: TeamGraphix#68). This commit adds two methods to the class `Pattern`: - `is_parameterized()` returns True if there is at least one measurement angle that is not just an instance of numbers.Number: indeed, a parameterized pattern is a pattern where at least one measurement angle is an expression that is not a number, typically an instance of `sympy.Expr` (but we don't force to choose sympy here). - `subs(variable, substitute)` returns a copy of the pattern where occurrences of the given variable in measurement angles are substituted by the given value. Substitution is performed by calling the method `subs` on measurement angles, if the method exists, which is the case in particular for `sympy.Expr`. If the substitution returns a number, this number is coerced to `float`, to get numbers that implement the full number protocol (in particular, sympy numbers don't implement `cos`).

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

This commit is yet another tentative to implement parameterized measurement patterns, to fulfill issue TeamGraphix#45 (previous tentative: TeamGraphix#68). This commit adds two methods to the class `Pattern`: - `is_parameterized()` returns True if there is at least one measurement angle that is not just an instance of numbers.Number: indeed, a parameterized pattern is a pattern where at least one measurement angle is an expression that is not a number, typically an instance of `sympy.Expr` (but we don't force to choose sympy here). - `subs(variable, substitute)` returns a copy of the pattern where occurrences of the given variable in measurement angles are substituted by the given value. Substitution is performed by calling the method `subs` on measurement angles, if the method exists, which is the case in particular for `sympy.Expr`. If the substitution returns a number, this number is coerced to `float`, to get numbers that implement the full number protocol (in particular, sympy numbers don't implement `cos`).

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

This commit is yet another tentative to implement parameterized measurement patterns, to fulfill issue TeamGraphix#45 (previous tentative: TeamGraphix#68). This commit adds two methods to the class `Pattern`: - `is_parameterized()` returns True if there is at least one measurement angle that is not just an instance of numbers.Number: indeed, a parameterized pattern is a pattern where at least one measurement angle is an expression that is not a number, typically an instance of `sympy.Expr` (but we don't force to choose sympy here). - `subs(variable, substitute)` returns a copy of the pattern where occurrences of the given variable in measurement angles are substituted by the given value. Substitution is performed by calling the method `subs` on measurement angles, if the method exists, which is the case in particular for `sympy.Expr`. If the substitution returns a number, this number is coerced to `float`, to get numbers that implement the full number protocol (in particular, sympy numbers don't implement `cos`).

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

shinich1 and others added 8 commits June 24, 2023 22:43

Add parameter class

63a676c

fix value binding

9e3dadb

version 0 of Paramter class

0caad92

completed Parameter class implementation

a844808

prepare branch for PR

c773fd2

reformat with black; add sympy reqruirements.txt

370a397

Merge branch '45-parameterized-measurement_patterns' into pr-branch-i…

606cc16

…ssue45

Merge branch '45-parameterized-measurement-patterns' into pr-branch-i…

7fe263b

…ssue45

shinich1 previously approved these changes Jul 6, 2023

View reviewed changes

shinich1 assigned pafloxy Jul 6, 2023

shinich1 mentioned this pull request Oct 25, 2023

Gflow finding algorithm #81

Merged

shinich1 deleted the branch TeamGraphix:45-parameterized-measurement-patterns May 8, 2024 14:28

shinich1 closed this May 8, 2024

benjvmin93 reviewed May 28, 2024

View reviewed changes

thierry-martinez mentioned this pull request May 30, 2024

Fix #45: Parameterized measurement patterns #158

Open

thierry-martinez added a commit to thierry-martinez/graphix that referenced this pull request May 31, 2024

Add parameter class and support for symbolic simulations

81bbeb7

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

thierry-martinez added a commit to thierry-martinez/graphix that referenced this pull request Jul 10, 2024

Add parameter class and support for symbolic simulations

e092e92

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

thierry-martinez added a commit to thierry-martinez/graphix that referenced this pull request Jul 25, 2024

Add parameter class and support for symbolic simulations

aa6d11a

`parameter.py` is inspired by TeamGraphix#68 but is split in Parameter and ParameterExpression.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Completed Parameter Implementation #68

Completed Parameter Implementation #68

pafloxy commented Jul 6, 2023

mgarnier59 commented Jul 7, 2023

shinich1 commented Jul 7, 2023

mgarnier59 commented Jul 7, 2023

shinich1 commented May 8, 2024 •

edited

Loading

pafloxy commented May 8, 2024

benjvmin93 May 28, 2024 •

edited

Loading

benjvmin93 May 28, 2024

benjvmin93 May 28, 2024

benjvmin93 May 28, 2024 •

edited

Loading

benjvmin93 May 28, 2024 •

edited

Loading

benjvmin93 May 28, 2024

Completed Parameter Implementation #68

Completed Parameter Implementation #68

Conversation

pafloxy commented Jul 6, 2023

mgarnier59 commented Jul 7, 2023

shinich1 commented Jul 7, 2023

mgarnier59 commented Jul 7, 2023

shinich1 commented May 8, 2024 • edited Loading

pafloxy commented May 8, 2024

benjvmin93 May 28, 2024 • edited Loading

Choose a reason for hiding this comment

benjvmin93 May 28, 2024

Choose a reason for hiding this comment

benjvmin93 May 28, 2024

Choose a reason for hiding this comment

benjvmin93 May 28, 2024 • edited Loading

Choose a reason for hiding this comment

benjvmin93 May 28, 2024 • edited Loading

Choose a reason for hiding this comment

benjvmin93 May 28, 2024

Choose a reason for hiding this comment

shinich1 commented May 8, 2024 •

edited

Loading

benjvmin93 May 28, 2024 •

edited

Loading

benjvmin93 May 28, 2024 •

edited

Loading

benjvmin93 May 28, 2024 •

edited

Loading