The Möbius function is defined as follows:
\[\begin{gather*} \forall x,y \in P: \quad \sum_{z \in [x,y]} \varphi(z,y) = \begin{cases} &1 \textit{, if } x=y \\ &0 \textit{, if } x \neq y \end{cases} \end{gather*}\]
By definition \(\varphi(12,12) = 1\) and
\[\begin{gather*} \varphi(6,12) + \varphi(12,12) = 0 \end{gather*}\]
meaning \(\varphi(6,12) = -1\). Thus, by
\[\begin{gather*} \varphi(3,12) + \varphi(6,12) + \varphi(12,12) = 0 \end{gather*}\]
we derive \(\varphi(3,12) = 0\). And by
\[\begin{gather*} \varphi(4,12) + \varphi(12,12) = 0 \end{gather*}\]
we get that \(\varphi(4,12) = -1\). Now we have:
\[\begin{gather*} \varphi(2,12) + \varphi(4,12) + \varphi(6,12) + \varphi(12,12) ) = 0 \end{gather*}\]
from which we can derive \(\varphi(2,12) = 1\) Therefore, we know by:
\[\begin{gather*} \varphi(1,12) + \varphi(2,12) + \varphi(3,12) + \varphi(4,12) + \varphi(6,12) + \varphi(12,12) = 0 \end{gather*}\]
that \(\varphi(1,12) = 0\). The last example from the lecture states:
\[\begin{gather*} \varphi(1,n) = \varphi(n)= \varphi_{\geq}(0,e_1) \cdots \varphi_{geq}(0,e_r) \end{gather*}\]
where \(n = p_1^{e_1} \cdots p_r^{e_r}\) (prime factorisation of \(n\)).
\[\begin{gather*} \varphi(n)= \begin{cases} 1 &\text{, if } n=1\\ (-1)^r &\text{, if } n=p_1 \cdots p_r\\ 0 &\text{, if $n$ is not square free}. \end{cases} \end{gather*}\]
Since \(1 \cdot 2 \cdot 3 \cdot 4 \cdot 6 \cdot 12\) is not square free \(\varphi(12)=0\), which coincides with our findings.
We know:
\[\begin{gather*} \varphi(4,4) = \varphi(3,3) = \varphi(2,2) = \varphi(1,1) = \varphi(0,0) = 1 \end{gather*}\]
and
\[\begin{gather*} \varphi(0,1) = \varphi(0,3) = \varphi(0,2) = \varphi(1,4) = \varphi(2,4) = \varphi(3,4) = -1 \end{gather*}\]
and by the calculated values and
\[\begin{gather*} \varphi(0,4) + \varphi(1,4) + \varphi(2,4) + \varphi(3,4) + \varphi(4,4) = 0 \end{gather*}\]
we derive \(\varphi(0,4) = 2\).
\[\begin{gather*} (a,x) \leq (b,x) \Leftrightarrow a \leq_1 b \land x \leq_2 y. \end{gather*}\]
Show that \((P,\leq)\) is a locally finite poset with \(0\)-element.
A poset has to satisfy reflexivity, asymmetry and transitivity. We first show that \(P\) is a poset.
For \(\varphi_P((a,x),(b,y))\) is has to hold that:
\[\begin{gather*} \sum_{(c,z) \in [(a,x),(b,y)]} \varphi_{P((c,z),(b,y))} = \begin{cases} 1 &\text{ ,if $a=b$ and $x=y$}\\ 0 &\text{ ,if $a \neq b$ or $x \neq y$} \end{cases} \end{gather*}\]
The first condition holds, since:
\(\varphi_{P_1}(a,b) = 1\), when \(a=b\) and \(varphi_{P_2}(x,y) = 1\) when \(x = y\) thus \(\varphi_P((a,x),(b,y)) = 1 \cdot 1 = 1\), if \(a=b\) and \(x = y\).
We show, the second condition holds:
\[\begin{gather*} \sum_{(c,z) \in [(a,x),(b,y)]} \varphi_P((c,z),(b,y)) = \sum_{(c,z) \in [(a,x),(b,y)]} \varphi_{P_1}(c,b) \cdot \varphi_{P_2}(z,y) = \end{gather*}\]
we split the sum
\[\begin{gather*} \sum_{(c) \in [a,b]} \sum_{(z) \in [x,y]} \varphi_{P_1}(c,b) \cdot \varphi_{P_2}(z,y) \end{gather*}\]
we extract the constant factor \(\varphi_{P_1}(c,b)\)
\[\begin{gather*} \sum_{(c) \in [a,b]} \varphi_{P_1}(c,b) \cdot \sum_{(z) \in [x,y]} \varphi_{P_2}(z,y) \end{gather*}\]
Now either \(x \neq y\) or \(a \neq b\), w.l.o.g. we say \(x = y\)
\[\begin{gather*} \sum_{(c) \in [a,b]} \varphi_{P_1}(c,b) \cdot 0 = 0 \end{gather*}\]
\[\begin{gather*} m - ((\frac{m}{p} + \frac{m}{q} + \frac{m}{r}) - (\frac{m}{pq} + \frac{m}{pr} + \frac{m}{qr}) + \frac{m}{pqr}) = \end{gather*}\]
since, \(m = pqr\)
\[\begin{gather*} = m - (\frac{pqr}{p} + \frac{pqr}{q} + \frac{pqr}{r} - \frac{pqr}{pq} - \frac{pqr}{pr} - \frac{pqr}{qr} + \frac{pqr}{pqr}) = \end{gather*}\]
we cancel the respective factors in each term
\[\begin{gather*} = m - qr - pr - pq + r + q + p - 1 \end{gather*}\]
Let \(d = gcd(a+b,a-b)\), then by \(\dagger\), \(d | (a+b)+(a-b)\), thus \(d|2a\), and \(d | (a+b)-(a-b)\), thus \(d | 2b\).
And since \(d|2a\) and \(d |2b\), \(d|gcd(2a,2b)\), by definition of the \(gcd\), and by \(gcd(a,b) = 1\), \(d\) can only either be \(1\) or \(2\).
Lemma \(\dagger\): Let \(z = gcd(x,y)\), then \(\exists m,n\), such that \(x=zm\) and \(y=zn\). Then \(x+y = zm + zn = z(m+n)\), thus \(z|(x+y)\) for arbitrary \(x,y\), where \(z = gcd(x,y)\) and \(x-y = zm - zn = z(m-n)\), thus \(z|(x-y)\) for arbitrary \(x,y\) such that \(z=gcd(x,y)\).
We know that forall positive integers, \(1 | n\), thus by \(\dagger\) \(2 | n\), therefore also \(3|n\). Since \(2\) and \(3\) divide \(n\), \(6 | n\), hence by \(\dagger\), \(7 |n\). Then, \(n = 2 \cdot 3 \cdot 7 = 42\).
\[\begin{gather*} 2863a + 1057b = 42 \end{gather*}\]
We first compute the division chain:
\[\begin{align} 2863 &= 1057 \cdot 2 + 749 \\ 1057 &= 749 \cdot 1 + 308 \\ 749 &=308 \cdot 2 + 133 \\ 308 &= 133 \cdot 2 + 42 \\ 133 &= 42 \cdot 3 + 7 \\ 42 &= 7 \cdot 6 + 0 \end{align}\]
We then find the linear combination for \(2863\) and \(1057\):
\[\begin{align} 42 &= 308 - (133 \cdot 2) \\ 42 &= 308 - ((749 - 308 \cdot 2) \cdot 2) = 308 \cdot 5 - 749 \cdot 2 \\ 42 &= (1057 - 749 \cdot 1) \cdot 5 - 749 \cdot 2 = 1057 \cdot 5 - 749 \cdot 7 \\ 42 &= 1057 \cdot 5 - (2863 - 1057 \cdot 2) \cdot 7 = 1057 \cdot 19 - 2863 \cdot 7 \end{align}\]
Thus we have found \(a = -7\) and \(b = 19\) which satisfy the equation.
\[\begin{align} x^3 + 5x^2 + 7x + 3 &= 1 \cdot (x^3 + x^2 - 5x + 3) + (4x^2 + 12x) \\ x^3 + x^2 - 5x + 3 &= (4x^2 + 12x) \cdot (1/4 x - 1/2) + (x+3) \\ 4x^2 + 12x &= (x+3) \cdot (4x) + 0 \end{align}\]
Thus, forall \(x \in \mathbb{Q}^x: a \cdot x + 3 \cdot a\) is a gcd of \(x^3 + 5x^2 + 7x + 3\) and \(x^3 + x^2 - 5x +3\).