Main

## Main.ObjectCspline History

February 17, 2018, at 05:55 AM by 184.254.42.171 -
February 17, 2018, at 05:54 AM by 184.254.42.171 -

(:toggle hide gekko button show="Example GEKKO (Python) Code":) (:div id=gekko:) (:source lang=python:) from gekko import gekko import numpy as np import matplotlib.pyplot as plt

""" minimize y s.t. y = f(x)

using cubic spline with random sampling of data """

1. Function to generate data for cspline

def f(x):

    return 3*np.sin(x) - (x-3)

1. Create model

c = gekko()

1. Cubic spline

x = c.Var(value=15) y = c.Var() x_data = np.random.rand(50)*10+10 y_data = f(x_data) c.cspline(x,y,x_data,y_data,True) c.Obj(y)

1. Options

c.options.IMODE = 3 c.options.CSV_READ = 0 c.options.SOLVER = 3 c.solve()

1. Generate continuous trend for plot

z = np.linspace(10,20,100)

1. Check if solved successfully

if c.options.SOLVESTATUS == 1:

    plt.figure()
plt.plot(z,f(z),'r-',label='original')
plt.scatter(x_data,y_data,5,'b',label='data')
plt.scatter(x.value,y.value,200,'k','x',label='minimum')
plt.legend(loc='best')


else:

    print ('Failed to converge!')
plt.figure()
plt.plot(z,f(z),'r-',label='original')
plt.scatter(x_data,y_data,5,'b')
plt.legend(loc='best')


plt.show() (:sourceend:) (:divend:)

February 14, 2018, at 04:29 PM by 174.148.12.56 -

(:html:) <iframe width="560" height="315" src="https://www.youtube.com/embed/s1jSLpDXvzs" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe> (:htmlend:)

February 14, 2018, at 12:26 AM by 174.148.43.131 -
Changed line 23 from:
to:
February 14, 2018, at 12:26 AM by 174.148.43.131 -
Changed lines 21-22 from:

The function is evaluated at the points x_data = [-1.0 -0.8 -0.5 -0.25 0.0 0.1 0.2 0.5].

to:

The function is evaluated at the points x_data = [-1.0 -0.8 -0.5 -0.25 0.0 0.1 0.2 0.5]. Evaluating at additional points shows the cubic spline interpolation function. The maximum of the original function is at x=0 with a result y=1. Because the cubic spline has only 8 points, there is some approximation error and the optimal solution of the cubic spline is slightly to the left of the true solution.

Changed line 25 from:

A cubic spline intersects the points to create the function approximations in the range of x between -1.0 and 0.5. There is extrapolation error outside of this range, as expected. Bounds on x should be added or additional cubic spline sample points should be added to avoid problems with optimizer performance in the extrapolation region.

to:

The cubic spline intersects the points to create the function approximations in the range of x between -1.0 and 0.5. There is extrapolation error outside of this range, as expected. Bounds on x should be added or additional cubic spline sample points should be added to avoid problems with optimizer performance in the extrapolation region.

February 14, 2018, at 12:15 AM by 174.148.43.131 -

Find the maximum of a function defined by 8 points that approximate the true function.

$$y(x) = \frac{1}{1+25 x^2}$$

The function is evaluated at the points x_data = [-1.0 -0.8 -0.5 -0.25 0.0 0.1 0.2 0.5].

A cubic spline intersects the points to create the function approximations in the range of x between -1.0 and 0.5. There is extrapolation error outside of this range, as expected. Bounds on x should be added or additional cubic spline sample points should be added to avoid problems with optimizer performance in the extrapolation region.

February 14, 2018, at 12:08 AM by 174.148.43.131 -
February 13, 2018, at 09:48 PM by 173.117.238.26 -
Changed lines 17-26 from:
 Objects

to:

(:source lang=python:) import numpy as np import matplotlib.pyplot as plt from APMonitor.apm import *

s = 'http://byu.apmonitor.com' a = 'cspline'

model = ''' Objects

Changed lines 28-30 from:
 End Objects

File c.csv

to:

End Objects

File c.csv

Changed lines 40-42 from:
 End File

Connections

to:

End File

Connections

Changed lines 45-47 from:
 End Connections

Variables

to:

End Connections

Parameters End Parameters

Variables

Changed lines 53-55 from:
 End Variables

Equations

to:

End Variables

Equations

Changed lines 57-80 from:
 End Equations

to:

End Equations '''

1. write file

fid = open('model.apm','w') fid.write(model) fid.close()

1. clear prior, load new model

1. set steady state optimiation and solve

apm_option(s,a,'apm.imode',3) output = apm(s,a,'solve') print(output)

1. retrieve solution

z = apm_sol(s,a)

1. print solution

print('x: ' + str(z['x'])) print('y: ' + str(z['y'])) (:sourceend:)

(:title Cubic Spline (cspline) Object:) (:keywords Cubic spline, Object, APMonitor, Option, Configure, Default, Description:) (:description One dimensional cubic spline for nonlinear function approximation with multiple interpolating functions that have continuous first and second derivatives:)

 Type: Object
Data: Two data vectors that define 1D function points
Inputs: Name of first data column (e.g. x)
Outputs: Name of second data column (e.g. y)
Description: Cubic spline for nonlinear function approximation


A cubic spline is a nonlinear function constructed of multiple third-order polynomials. These polynomials pass through a set of control points and have continuous first and second derivatives everywhere. The second derivative is set to zero at the left and right endpoints, to provide a boundary condition to complete the system of equations. There is poor extrapolation when function retrievals are requested outside of the data points. The input should be constrained or else additional data points added to avoid extrapolation.

Example Usage

 Objects
c = cspline
End Objects

File c.csv
x_data         ,  y_data
-1.0000000e+00 ,  3.8461538e-02
-8.0000000e-01 ,  5.8823529e-02
-5.0000000e-01 ,  1.3793103e-01
-2.5000000e-01 ,  3.9024390e-01
0.0000000e+00 ,  1.0000000e+00
1.0000000e-01 ,  8.0000000e-01
2.0000000e-01 ,  5.0000000e-01
5.0000000e-01 ,  1.3793103e-01
End File

Connections
x = c.x_data
y = c.y_data
End Connections

Variables
x = -0.5  >= -1 <= 0.5
y
End Variables

Equations
maximize y
End Equations