Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I would like to know what is known (both explanations and references) about the spaces of smooth solutions to linear systems of PDEs of the following form:

Let $g_{1},...,g_{n}$ be smooth functions on $\mathbb{R}^{n}$ with the integrability condition $\partial{g_{i}}/\partial{x^{j}}=\partial{g_{j}}/\partial{x^{i}}$ and consider the space of smooth functions $f$ on $\mathbb{R}^{n}$ satisfying $\partial{f}/\partial{x^{i}}=fg_{i}$ for all $i$.

Similarly for the $g_{i}$ and $f$ being holomorphic on $\mathbb{C}^{n}$, and replacing $\mathbb{R}^{n}, \mathbb{C}^{n}$ with open contractible subsets.

My hope is that the answer is there is a unique solution, up to scaling.

share|cite|improve this question
You can rewrite your system as $\vec{\nabla} \ln f = \vec{g}$... – anon Jul 29 '11 at 17:20
Does that help? – Pascal Jul 29 '11 at 17:22
Also, what if $f$ has zeroes and what if I'm interested in complex values? I would be happy to say that there exists a contractible open cover of my space over which there is a unique solution on each covering set, if that would help solve the problem about branches of the logarithm. – Pascal Jul 29 '11 at 17:31
Note that $f$ can only have zeros where $\vec{g}$ has singularities. Remember that $\vec{g}$ is the given and $f$ the unknown, so that when looking at the latter in terms of the former we are working with exponentials instead of logarithms and therefore don't have to worry about branches. – anon Jul 29 '11 at 17:38

You can rewrite the system of PDEs as $\vec{\nabla} \ln f = \vec{g}$. If the vector field $\vec{g}$ has a potential function (which I believe is true if and only if its Jacobian matrix is symmetric, but I don't recall the source of this fact off-hand), then we may denote the potential as $G$ and solve the system as $f=\exp G$, which is unique up to rescalings. (There may be a negative sign in the exponential depending on what definition of potential you're using.) On the other hand, if $\vec{g}$ doesn't have a potential then it isn't a gradient field and hence no $f$ exists.

share|cite|improve this answer

anon's answer is almost complete, here I'm filling important gaps.

As anon pointed your system is $\frac{\partial}{\partial x_i} \ln(f)= g_i$ then by Froebenius theorem, the PDE system has a solution if and only if the $\nabla g$ is a symmetric matrix, which is true given your assumption that $\frac{\partial}{\partial x_i} g_j= \frac{\partial}{\partial x_j} g_j$.

Moreover, given initial data, Frobenius theorem also gives you a unique local solution for you system, see page 92 of this paper (‎) for the Frobenius theorem in a PDE/local coordinates/classical version. Perhaps with your additional assumptions about the domain, you maybe able to extent it to a global solution.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.