Difference between revisions of "SymPy/Differential Equations/RC Example"

From PrattWiki
Jump to navigation Jump to search
(Created page with "== Description == The following page is adapted from Maple/Differential Equations/RC Example and will go through an example of using SymPy's ability to symbolically solve...")
 
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
== Description ==
 
== Description ==
 
The following page is adapted from [[Maple/Differential Equations/RC Example]] and will go through an example of using SymPy's ability to symbolically solve a differential equation - specifically to analyze a circuit that undergoes a change in source values.  In this particular case, the independent source is given as a constant for all times before 0 sec, at which time it changes to a non-constant source.  The source is connected to a network containing both resistors and a capacitor. While there are several ways to put all this work together in Python, the following will provide a flexible framework for solving the equations and using substitution to determine a numerical list of initial conditions, substituting element values into the differential equations and putting them into a list, solving the differential equations, and finally plotting the results.   
 
The following page is adapted from [[Maple/Differential Equations/RC Example]] and will go through an example of using SymPy's ability to symbolically solve a differential equation - specifically to analyze a circuit that undergoes a change in source values.  In this particular case, the independent source is given as a constant for all times before 0 sec, at which time it changes to a non-constant source.  The source is connected to a network containing both resistors and a capacitor. While there are several ways to put all this work together in Python, the following will provide a flexible framework for solving the equations and using substitution to determine a numerical list of initial conditions, substituting element values into the differential equations and putting them into a list, solving the differential equations, and finally plotting the results.   
 +
 +
'''''Note:''''' at the moment, SymPy's <code>dsolve</code> command works pretty well for up to two coupled ODEs but starts having issues once a third equation is in the mix.  Unlike [[Maple]], which has the option to use Laplace techniques with its <code>dsolve</code> command, SymPy does not.  Roundoff and other errors creep in, causing the solution to fail to match the initial conditions.  For higher-order systems, at the moment, a numerical solver such as <code>solve_ivp</code> might be required (which also requires breaking the system down into a series of first-order differential equations).
 +
 +
The entire notebook is available via the following link to [https://colab.research.google.com/drive/17Y-DpLVSSn5um4v3w0MoH2UzRD1v90T4?usp=sharing IVP_RCdemo.ipynb]
  
 
== Circuit ==
 
== Circuit ==
Line 38: Line 42:
  
 
== Code ==
 
== Code ==
Now that the equations for DC steady state and for the differential model are known, you can write Maple code to solve for everything.
+
Now that the equations for DC steady state and for the differential model are known, you can write a Python script to solve for everything.
 +
 
 +
=== Initialization ===
 +
You will need to import <code>sympy</code> to use symbolic calculations.  You will also need <code>numpy</code> and <code>matplotlib.pyplot</code> to make a graph.
 +
<syntaxhighlight lang=python>
 +
import sympy as sym
 +
import numpy as np
 +
import matplotlib.pyplot as plt
 +
</syntaxhighlight>
 +
 
 +
Because you are solving systems of differential equations, the same helper functions that were used to solve [[SymPy/Simultaneous_Equations|simultaneous equations]] will be useful.  Given that, your script should import the helper functions from [https://tinyurl.com/mrgdemos/SymPy_Files/sym_helper.py <code>sym_helper.py</code>]; you can do this with:
 +
<syntaxhighlight lang=python>
 +
# Import the helper functions
 +
try:
 +
  from sym_helper import *
 +
except: # Get sym_helper.py from GitHub
 +
  import requests
 +
  url = 'https://tinyurl.com/mrgdemos/SymPy_Files/sym_helper.py'
 +
  open('sym_helper.py', 'w').write(requests.get(url).text)
 +
  from sym_helper import *
 +
</syntaxhighlight>
 +
 
 +
Finally, you will want the script to use $$\LaTeX$$ to display your equations if possible:
 +
<syntaxhighlight lang=python>
 +
# Use LaTeX if possible
 +
sym.init_printing(use_latex=True)
 +
</syntaxhighlight>
 +
 
 +
=== Declaring Variables and Functions ===
 +
Next you should tell SymPy what your variables are.  You will also need to declare the ''functions' that you will be using later.  For this example, the known values include the two resistors, the capacitor, the DC value of the source before time 0, and time.  The functions you will want to declare include the (potentially non-constant) voltage of the source after time 0 and the voltage across the capacitor.  You may also want to create a list of the known variables and a list of the values of those variables:
 +
<syntaxhighlight lang=python>
 +
R1, R2, C, vsdc, t = sym.var('R1, R2, C, v_{s\,dc}, t')
 +
vs = sym.Function("v_s")
 +
vC = sym.Function("v_C")
 +
knowns = R1, R2, C, vsdc, vs(t)
 +
knownvals = [20000, 30000, 50e-6, 5, 10*sym.cos(8*t)]
 +
</syntaxhighlight>
  
=== Preparing the Worksheet ===
 
Be sure that your name and the assignment show up as text at the top
 
of the page.  Also be sure that the first Maple command is <code>restart</code>.
 
  
 
=== Initial Conditions From Steady-State ===
 
=== Initial Conditions From Steady-State ===
While Maple can solve differential equations with symbolic initial
+
While SymPy can solve some differential equations with symbolic initial conditions and coefficients, most of the time this will result in a very unwieldy and unhelpful representation.  For this assignment, you will be providing SymPy with numerical values for the initial conditions.  To do this, we are going to ask SymPy to solve the linear algebra problem that comes from the DC steady state and then we will substitute in numerical values.  First, we will define a variable to be the initial condition of the capacitor voltage and will add it to a list of unknowns:
conditions and coefficients, most of the
+
<syntaxhighlight lang=python>
time this will result in a very unwieldy and unhelpful
+
vC0 = vC(t).subs(t, 0)
representation.  For this assignment, you will be providing Maple with
+
unksSS = [vC0]
numerical values for the initial conditions.  The three step process
+
</syntaxhighlight>
for this is to set up equations for the DC steady-state values in terms
+
Next, we will write the conservation equation we have at DC steady-state and add it to a list of equations:
of the sources and elements:
+
<syntaxhighlight lang=python>
sseqn1 := (v__C(0)-v__S,DC)/R__1 + v__C(0)/R__2 = 0
+
eqn1SS = (vC0-vsdc) / R1 + vC0 / R2
solve those equations,
+
eqnsSS = [eqn1SS]
sssoln := solve({sseqn1}, [v__C(0)])
+
</syntaxhighlight>
and then later you will substitute in the proper element and source values. 
+
Finally, we can solve the equation and then substitute in numerical values:
The output of this should be a set of equations, where where the set is surrounded by double square brackets.
+
<syntaxhighlight lang=python>
 +
solnSS = sym.solve(eqnsSS, unksSS)
 +
solnSSnum = makesubs(solnSS, knowns, knownvals)
 +
</syntaxhighlight>
 +
This results in <code>solnSSnum</code> being a dictionary with the initial value of $$v_C$$ defined in it, $$\{v_C(0):3\}$$
  
 
=== Differential Equations ===
 
=== Differential Equations ===
Next set up the differential equations, generate a single list of
+
Next we can set up and solve the differential equations.  First, make a list of all the unknown functions for which you want to solve:
the differential equations with numerical values substituted in for
+
<syntaxhighlight lang=python>
the elements and sources, and solve.  This is a three step process: define the
+
unks = [vC(t)]
equations:
+
</syntaxhighlight>
deqn1 := (v__C(t)-v__S(t))/R__1+v__C(t)/R__2+C*(diff(v__C(t), t)) = 0
 
then define and substitute in the element and source values to your initial value solutions and your differential equations:
 
vals := R__1 = 20000., R__2 = 30000., C = 0.0000500, v__S,DC = 5.0, v__S(t) = 10.0*cos(8*t)
 
numeqn := subs(vals, {deqn1, sssoln[][]})
 
The end result will be a set of equations with numbers instead of symbols (except for the variables of interest).  Now solve the equations:
 
soln := dsolve(numeqn, [v__C(t)])
 
You may want to look at a simplified version of the solution by first converting it to
 
cos, sin, and exponentials (in case there are hyperbolic trig functions) and allowing
 
Maple to expand and combine terms, then round off to four significant figures:
 
evalf[4](combine(expand(convert(soln, expsincos))))
 
  
Note that in some cases the results of the differential equation are,
+
You can now write the differential equations (using <code>sym.Derivative(FUN, VAR)</code> to represent the derivatives) and put them in a list:
frankly, ugly. Sometimes, telling Maple to solve using the Laplace
+
<syntaxhighlight lang=python>
method comes up with a more compact answer:
+
eq1 = (vC(t)-vs(t))/R1 + vC(t)/R2 + C*sym.Derivative(vC(t), t)
soln := dsolve(numeqn, [v__C(t)], method=laplace):
+
eqns = [eq1]
evalf[4](combine(expand(convert(soln, expsincos))))
+
</syntaxhighlight>
Other times, neither Maple's default method nor Laplace have a
+
 
``nice'' answer; in those cases, simply put a colon at the end of the
+
While it is possible for SymPy to solve some systems purely symbolically, there are times where it is much easier for it to solve equations where the coefficients are numbers.  To help with that, the <code>sym_helper</code> file has a command called <code>makesubseqnlist(LIST, KNOWNVARLIST, KNOWNVALLIST)</code> that will take a list of equations and substitute the given values in for the given variables.  The code for <code>makesubeqnlist</code> is:
line to suppress the output and forget about the simplify line:
+
<syntaxhighlight lang=python>
soln := dsolve(numeqn, [v__C(t)]):
+
def makesubseqnlist(equations, vartosub, valtosub):
In those cases, you will want to focus more on the plot than the
+
    sublist = list(zip(vartosub, valtosub))
analytical solution.
+
    esub = []   
 +
    for eqn in equations:
 +
      esub.append(eqn.subs(sublist))
 +
    return esub
 +
</syntaxhighlight>
 +
and the code that will go in your script for this is:
 +
<syntaxhighlight lang=python>
 +
neqns = makesubseqnlist(eqns, knowns, knownvals)
 +
</syntaxhighlight>
 +
 
 +
Now you can solve the differential equations using <code>dsolve</code>:
 +
<syntaxhighlight lang=python>
 +
solnlist = sym.dsolve(neqns, unks, ics=solnSSnum)
 +
</syntaxhighlight>
 +
 
 +
'''''Note:''''' unlike <code>solve</code>, which creates a dictionary of solutions, <code>dsolve</code> creates a list.  It turns out having a dictionary makes life much easier later, so <code>sym_helper</code> has a <code>make_dict(LIST)</code> command that will turn a list of equations into a dictionary.  The code for <code>make_dict</code> is:
 +
<syntaxhighlight lang=python>
 +
def makedict(sollist):
 +
    d = {}
 +
    for eqn in sollist:
 +
        d[eqn.lhs] = eqn.rhs
 +
    return d
 +
</syntaxhighlight>
 +
and so the code you can add to your script is:
 +
<syntaxhighlight lang=python>
 +
def makedict(sollist):
 +
soln = makedict(solnlist)
 +
</syntaxhighlight>
  
 
=== Plotting ===
 
=== Plotting ===
Plotting can sometimes be a little more complicated than it seems - much of the
+
Plotting will require using <code>lambdify</code> to turn one or more functions into a lambda function and then setting up your plotHere is the code for this example that does all those things (making a plot that runs from $$t$$=0 to 10 sec):
time, round-off errors will cause solutions that have tiny vestigial
+
<syntaxhighlight lang=python>
imaginary values. To eliminate this, you can have Maple <code>map</code> the
+
vCfun = sym.lambdify(t, soln[vC(t)])
real part of the solution vectorThat is:
+
fig = plt.figure(num=1, clear=True)
plot(map(Re, subs(soln, [v__C(t)])), t = 0 .. 10,  
+
ax = fig.add_subplot(1, 1, 1)
labels = [typeset(t, ", s"), typeset(v__C(t), ", V")])
+
tn = np.linspace(0, 10, 1000)
 +
ax.plot(tn, vCfun(tn))
 +
</syntaxhighlight>
 +
 
  
 
== Complete Example ==
 
== Complete Example ==
For the Spring 2022 semester, students in EGR 224 get the code by going to the [https://duke.box.com/s/mkohvjf6fvt2w6nxcbz1ou21ghyfz4gz public Box folder for EGR 224] and looking at the Lab05files folderIn addition to the worksheet for the circuit above, there are worksheets for several example and practice problems from Alexander & Sadiku 7e Chapters 7 and 8.
+
For the Spring 2023 semester, students in EGR 224 get the code by going to the [https://colab.research.google.com/drive/17Y-DpLVSSn5um4v3w0MoH2UzRD1v90T4?usp=sharing Google Colab] version of itThis notebook includes several more comments than are listed above and has multiple instances where it displays intermediate results.

Latest revision as of 17:18, 19 February 2023

Description

The following page is adapted from Maple/Differential Equations/RC Example and will go through an example of using SymPy's ability to symbolically solve a differential equation - specifically to analyze a circuit that undergoes a change in source values. In this particular case, the independent source is given as a constant for all times before 0 sec, at which time it changes to a non-constant source. The source is connected to a network containing both resistors and a capacitor. While there are several ways to put all this work together in Python, the following will provide a flexible framework for solving the equations and using substitution to determine a numerical list of initial conditions, substituting element values into the differential equations and putting them into a list, solving the differential equations, and finally plotting the results.

Note: at the moment, SymPy's dsolve command works pretty well for up to two coupled ODEs but starts having issues once a third equation is in the mix. Unlike Maple, which has the option to use Laplace techniques with its dsolve command, SymPy does not. Roundoff and other errors creep in, causing the solution to fail to match the initial conditions. For higher-order systems, at the moment, a numerical solver such as solve_ivp might be required (which also requires breaking the system down into a series of first-order differential equations).

The entire notebook is available via the following link to IVP_RCdemo.ipynb

Circuit

For this demonstration code, the following circuit is used:

RCD DemoCircuit.png

where \(R_1\)=20 k\(\Omega\), \(R_2\)=30 k\(\Omega\), \(C\)=50 \(\mu\)F, and \(v_s(t)\) changes from 5 V to 10 cos(8\(t\)) V when \(t=0\) s.

DC Steady-State Analysis

Assuming the circuit has been in place for a "long time" before \(t\)=0 sec, and given the topology of the circuit and the fact that the independent source is a constant for all times before 0 sec, you can use the DC steady-state equivalent for the circuit at \(t=0^-\) sec:

RCD DemoCircuitSS.png

Using KCL at the top right node gives:

\( \frac{v_C(0^-)-v_S}{R_1}+ \frac{v_C(0^-)}{R_2}=0\! \)

which can be solved to find the capacitor voltage at the time just before the source \(v_S(t)\) changes.

Model Equations for t>0

In general, after \(t\)=0 sec you can label the circuit as:

RCD DemoCircuitDE.png

Using KCL at the top right node again gives:

\( \frac{v_C(t)-v_S(t)}{R_1}+ \frac{v_C(t)}{R_2}+C\frac{dv_C(t)}{dt}=0\! \)


Code

Now that the equations for DC steady state and for the differential model are known, you can write a Python script to solve for everything.

Initialization

You will need to import sympy to use symbolic calculations. You will also need numpy and matplotlib.pyplot to make a graph.

import sympy as sym
import numpy as np
import matplotlib.pyplot as plt

Because you are solving systems of differential equations, the same helper functions that were used to solve simultaneous equations will be useful. Given that, your script should import the helper functions from sym_helper.py; you can do this with:

# Import the helper functions
try:
  from sym_helper import *
except: # Get sym_helper.py from GitHub
  import requests
  url = 'https://tinyurl.com/mrgdemos/SymPy_Files/sym_helper.py'
  open('sym_helper.py', 'w').write(requests.get(url).text)
  from sym_helper import *

Finally, you will want the script to use $$\LaTeX$$ to display your equations if possible:

# Use LaTeX if possible
sym.init_printing(use_latex=True)

Declaring Variables and Functions

Next you should tell SymPy what your variables are. You will also need to declare the functions' that you will be using later. For this example, the known values include the two resistors, the capacitor, the DC value of the source before time 0, and time. The functions you will want to declare include the (potentially non-constant) voltage of the source after time 0 and the voltage across the capacitor. You may also want to create a list of the known variables and a list of the values of those variables:

R1, R2, C, vsdc, t = sym.var('R1, R2, C, v_{s\,dc}, t')
vs = sym.Function("v_s")
vC = sym.Function("v_C")
knowns = R1, R2, C, vsdc, vs(t)
knownvals = [20000, 30000, 50e-6, 5, 10*sym.cos(8*t)]


Initial Conditions From Steady-State

While SymPy can solve some differential equations with symbolic initial conditions and coefficients, most of the time this will result in a very unwieldy and unhelpful representation. For this assignment, you will be providing SymPy with numerical values for the initial conditions. To do this, we are going to ask SymPy to solve the linear algebra problem that comes from the DC steady state and then we will substitute in numerical values. First, we will define a variable to be the initial condition of the capacitor voltage and will add it to a list of unknowns:

vC0 = vC(t).subs(t, 0)
unksSS = [vC0]

Next, we will write the conservation equation we have at DC steady-state and add it to a list of equations:

eqn1SS = (vC0-vsdc) / R1 + vC0 / R2
eqnsSS = [eqn1SS]

Finally, we can solve the equation and then substitute in numerical values:

solnSS = sym.solve(eqnsSS, unksSS)
solnSSnum = makesubs(solnSS, knowns, knownvals)

This results in solnSSnum being a dictionary with the initial value of $$v_C$$ defined in it, $$\{v_C(0):3\}$$

Differential Equations

Next we can set up and solve the differential equations. First, make a list of all the unknown functions for which you want to solve:

unks = [vC(t)]

You can now write the differential equations (using sym.Derivative(FUN, VAR) to represent the derivatives) and put them in a list:

eq1 = (vC(t)-vs(t))/R1 + vC(t)/R2 + C*sym.Derivative(vC(t), t)
eqns = [eq1]

While it is possible for SymPy to solve some systems purely symbolically, there are times where it is much easier for it to solve equations where the coefficients are numbers. To help with that, the sym_helper file has a command called makesubseqnlist(LIST, KNOWNVARLIST, KNOWNVALLIST) that will take a list of equations and substitute the given values in for the given variables. The code for makesubeqnlist is:

def makesubseqnlist(equations, vartosub, valtosub):
    sublist = list(zip(vartosub, valtosub))
    esub = []    
    for eqn in equations:
      esub.append(eqn.subs(sublist))
    return esub

and the code that will go in your script for this is:

neqns = makesubseqnlist(eqns, knowns, knownvals)

Now you can solve the differential equations using dsolve:

solnlist = sym.dsolve(neqns, unks, ics=solnSSnum)

Note: unlike solve, which creates a dictionary of solutions, dsolve creates a list. It turns out having a dictionary makes life much easier later, so sym_helper has a make_dict(LIST) command that will turn a list of equations into a dictionary. The code for make_dict is:

def makedict(sollist):
    d = {}
    for eqn in sollist:
        d[eqn.lhs] = eqn.rhs
    return d

and so the code you can add to your script is:

def makedict(sollist):
soln = makedict(solnlist)

Plotting

Plotting will require using lambdify to turn one or more functions into a lambda function and then setting up your plot. Here is the code for this example that does all those things (making a plot that runs from $$t$$=0 to 10 sec):

vCfun = sym.lambdify(t, soln[vC(t)])
fig = plt.figure(num=1, clear=True)
ax = fig.add_subplot(1, 1, 1)
tn = np.linspace(0, 10, 1000)
ax.plot(tn, vCfun(tn))


Complete Example

For the Spring 2023 semester, students in EGR 224 get the code by going to the Google Colab version of it. This notebook includes several more comments than are listed above and has multiple instances where it displays intermediate results.