# Check satisfaction of equilibrium conditions¶

If EQTK’s solve algorithm converges, equilibrium and conservation laws are guaranteed to be satisfied (see algorithmic details). Nonetheless, it is sometimes useful to check if equilibrium conditions are met. To understand the discussion below, it may help to refresh yourself of the core concepts behind the couple equilibrium problem.

As a simple example, consider the chemical reactions

AB ⇌ A + B ; K = K₁

AC ⇌ A + C ; K = K₂

The equilibrium conditions are

$\begin{split}&K_1 = \frac{c_\mathrm{A}\,c_\mathrm{B}}{c_\mathrm{AB}},\\ &K_2 = \frac{c_\mathrm{A}\,c_\mathrm{C}}{c_\mathrm{AC}}.\end{split}$

Additionally, three conservation laws, corresponding to conservation of A, B, and C, must be satisfied.

$\begin{split}&c_\mathrm{A}^0 = c_\mathrm{A} + c_\mathrm{AB} + c_\mathrm{AC},\\ &c_\mathrm{B}^0 = c_\mathrm{B} + c_\mathrm{AB},\\ &c_\mathrm{C}^0 = c_\mathrm{C} + c_\mathrm{AC}.\end{split}$

Together, these five equations define the equilibrium. We can write these conditions in terms of the stoichiometric matrix $$\mathsf{N}$$ and conservation matrix $$\mathsf{A}$$. The stoichiometric matrix corresponding to the chemical reactions is

$\begin{split}\mathsf{N} = \begin{pmatrix} 1 & 1 & 0 & -1 & 0 \\ 1 & 0 & 1 & 0 & -1 \end{pmatrix}.\end{split}$

The conservation matrix $$\mathsf{A}$$ has rows that span the null space of $$\mathsf{N}$$.

$\begin{split}\mathsf{A} = \begin{pmatrix} 1 & 0 & 0 & 1 & 0 \\ 0 & 1 & 0 & 1 & 1 \\ 0 & 0 & 1 & 0 & 1 \end{pmatrix}.\end{split}$

If we define

$\begin{split}&\mathbf{c} = (c_\mathrm{A}, c_\mathrm{B}, c_\mathrm{C}, c_\mathrm{AB}, c_\mathrm{AC})^\mathsf{T},\\ &\mathbf{c}^0 = (c_\mathrm{A}^0, c_\mathrm{B}^0, c_\mathrm{C}^0, c_\mathrm{AB}^0, c_\mathrm{AC}^0)^\mathsf{T},\end{split}$

the equilibrium conditions are

$\begin{split}&K_1 = \prod_j c_j^{N_{1j}},\\ &K_2 = \prod_j c_j^{N_{2j}},\end{split}$

or more generally

$K_i = \prod_j c_j^{N_{ij}} \;\forall i.$

The condition for conservation is

$\mathsf{A}\cdot\mathbf{c} = \mathsf{A} \cdot \mathbf{c}^0.$

## Check of satisfaction of equilibrium and conservation conditions¶

The eqtk.eqcheck() function conveniently checks to make sure equilibrium and conservation conditions are met. To check equilibrium conditions, it verifies that

$\frac{\prod_j c_j^{N_{ij}}}{K_i} \approx 1 \;\forall i,$

where equality is checked to some tolerance.

Similarly to check conservation conditions, it verifies that

$\mathsf{A}\cdot\mathbf{c} - \mathsf{A} \cdot \mathbf{c}^0 \approx \mathbf{0}.$

Let’s use this function to verify that an equilibrium calculation was successful. To start with, we will use Numpy arrays as inputs. (In what follows, we assume EQTK, Numpy, and Pandas are all imported.)

c0 = [1, 0.5, 0.25, 0, 0]

N = [[1,  1,  0, -1,  0],
[1,  0,  1,  0, -1]]

K = [0.015, 0.003]

# Solve
c = eqtk.solve(c0=c0, N=N, K=K, units="mM")

# Verify calculation converged
eqtk.eqcheck(c, c0=c0, N=N, K=K, units="mM")


The last function call returns True.

If we instead N stored as a data frame and c0 as a series or data frame, it is not necessary to supply K, as it is already in the N data frame, nor is it necessary to supply c0 or units, as they can be inferred from c.

names = ["A", "B", "C", "AB", "AC"]
c0 = pd.Series(data=[1, 0.5, 0.25, 0, 0], index=names)

N = pd.DataFrame(data=[[1,  1,  0, -1,  0],
[1,  0,  1,  0, -1]],
columns=names)
N['equilibrium_constant'] = [0.015, 0.003]

# Solve
c = eqtk.solve(c0=c0, N=N, units="mM")

# Verify calculation converged
eqtk.eqcheck(c, N=N)


This calculation again returns True.

## Quantitative check in error in equilibrium and conservation conditions¶

To get more detailed information, specifically the value of the ratio

$\frac{\prod_j c_j^{N_{ij}}}{K_i}$

and the difference

$\mathsf{A}\cdot\mathbf{c} - \mathsf{A} \cdot \mathbf{c}^0,$

you can use the return_detailed=True keyword argument of eqtk.eqcheck(). With this keyword argument, it returns

1. A Boolean as to whether all equilibrium and conservation conditions are met.

2. An array of the ratios $$\prod_j c_j^{N_{ij}}/K_i$$.

3. An array of Booleans that are True if this ratio is close to unity.

4. An array of differences $$\mathbf{A}_i\cdot\mathbf{c} - \mathbf{A}_i \cdot \mathbf{c}^0$$ for each row $$i$$ in $$\mathsf{A}$$.

5. An array of Booleans that are True if this difference is close to zero.

Running

eqtk.eqcheck(c, N=N, return_detailed=True)


returns

(True,
array([1., 1.]),
array([ True,  True]),
array([6.17794068e-17, 3.43217750e-17, 1.57451566e-16]),
array([ True,  True,  True]))