Constrained tuning: Difference between revisions

Line 35:

== Computation ==

As a standard optimization problem, numerous algorithms exist to solve it, such as {{w|sequential quadratic programming}}, to name one. [[Flora Canou]]'s [https://github.com/FloraCanou/temperament_evaluator/blob/~~v0.26.4~~/te_optimizer_legacy.py tuning optimizer] ~~is such an implementation~~ in [https://www.python.org Python]~~. Note: it uses~~ [https://scipy.org/ Scipy].

As a standard optimization problem, numerous algorithms exist to solve it, such as {{w|sequential quadratic programming}}, to name one. [[Flora Canou]]'s [https://github.com/FloraCanou/temperament_evaluator/blob/master/te_optimizer_legacy.py tuning optimizer] solves constrained tuning problems in [https://www.python.org Python], using [https://scipy.org/ Scipy].

{{Databox| Code |

# This work is licensed under the GNU General Public License version 3.

# Version 0.28.0

import warnings

import numpy as np

from scipy import optimize, linalg

np.set_printoptions (suppress = True, linewidth = 256, precision = 4)

np.set_printoptions (suppress = True, linewidth = 256, precision = 3)

PRIME_LIST = [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89]

PRIME_LIST = [

2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37,

41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89,

]

class SCALAR:

Line 115:

Line 119:

print ("Euclidean tuning without constraints, solved using lstsq. ")

else:

gen0 = [SCALAR.CENT]*breeds.shape[0] #initial guess

gen0 = just_tuning_map[:breeds.shape[0]] #initial guess

cons = () ~~if cons_monzo_list is None else {~~

cons = optimize.LinearConstraint ((breeds @ cons_monzo_list).T,

~~'type': 'eq'~~,

lb = (just_tuning_map @ cons_monzo_list).T,

~~'fun': lambda gen:~~ (~~gen @ breeds -~~ just_tuning_map) @ cons_monzo_list

ub = (just_tuning_map @ cons_monzo_list).T

}

)

res = optimize.minimize (lambda gen: linalg.norm (gen @ breeds_x - just_tuning_map_x, ord = norm.order), gen0,

method = "~~SLSQP~~"~~, options = {'ftol': 1e-9}~~, constraints = cons)

method = "COBYQA", constraints = cons)

print (res.message)

if res.success:

Line 161:

Line 165:

<pre>

~~Optimization terminated successfully.~~

The lower bound for the trust-region radius has been reached

Generators: [1200. 1896.~~9521~~] (¢)

Generators: [1200. 1896.952] (¢)

Tuning map: [1200. 1896.~~9521~~ 2787.~~8085~~ 3369.~~5214~~] (¢)

Tuning map: [1200. 1896.952 2787.809 3369.521] (¢)

Error map: [ 0. -5.~~0029~~ 1.~~4948~~ 0.~~6955~~] (¢)

Error map: [ 0. -5.003 1.495 0.695] (¢)

</pre>

@@ Line 35: / Line 35: @@
 == Computation ==
-As a standard optimization problem, numerous algorithms exist to solve it, such as {{w|sequential quadratic programming}}, to name one. [[Flora Canou]]'s [https://github.com/FloraCanou/temperament_evaluator/blob/v0.26.4/te_optimizer_legacy.py tuning optimizer] is such an implementation in [https://www.python.org Python]. Note: it uses [https://scipy.org/ Scipy].
+As a standard optimization problem, numerous algorithms exist to solve it, such as {{w|sequential quadratic programming}}, to name one. [[Flora Canou]]'s [https://github.com/FloraCanou/temperament_evaluator/blob/master/te_optimizer_legacy.py tuning optimizer] solves constrained tuning problems in [https://www.python.org Python], using [https://scipy.org/ Scipy].
 {{Databox| Code |
 <syntaxhighlight lang="python">
-# © 2020-2023 Flora Canou | Version 0.27.2
+# © 2020-2025 Flora Canou
 # This work is licensed under the GNU General Public License version 3.
+# Version 0.28.0
 import warnings
 import numpy as np
 from scipy import optimize, linalg
-np.set_printoptions (suppress = True, linewidth = 256, precision = 4)
+np.set_printoptions (suppress = True, linewidth = 256, precision = 3)
-PRIME_LIST = [2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89]
+PRIME_LIST = [
+, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37,
+, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89,
+]
 class SCALAR:
@@ Line 115: / Line 119: @@
          print ("Euclidean tuning without constraints, solved using lstsq. ")
      else:
-         gen0 = [SCALAR.CENT]*breeds.shape[0] #initial guess
+         gen0 = just_tuning_map[:breeds.shape[0]] #initial guess
-         cons = () if cons_monzo_list is None else {
+         cons = optimize.LinearConstraint ((breeds @ cons_monzo_list).T,
-             'type': 'eq',
+             lb = (just_tuning_map @ cons_monzo_list).T,
-             'fun': lambda gen: (gen @ breeds - just_tuning_map) @ cons_monzo_list
+             ub = (just_tuning_map @ cons_monzo_list).T
-         }
+         )
          res = optimize.minimize (lambda gen: linalg.norm (gen @ breeds_x - just_tuning_map_x, ord = norm.order), gen0,
-             method = "SLSQP", options = {'ftol': 1e-9}, constraints = cons)
+             method = "COBYQA", constraints = cons)
          print (res.message)
          if res.success:
@@ Line 161: / Line 165: @@
 <pre>
-Optimization terminated successfully.
+The lower bound for the trust-region radius has been reached
-Generators: [1200.     1896.9521] (¢)
+Generators: [1200.    1896.952] (¢)
-Tuning map: [1200.     1896.9521 2787.8085 3369.5214] (¢)
+Tuning map: [1200.    1896.952 2787.809 3369.521] (¢)
-Error map: [ 0.     -5.0029  1.4948  0.6955] (¢)
+Error map: [ 0.    -5.003  1.495  0.695] (¢)
 </pre>