# OutOfMemoryError in HomotopyContinuation

**URL:** <https://discourse.julialang.org/t/outofmemoryerror-in-homotopycontinuation/95105>\
**Category:** General Usage\
**Tags:** question, package, memory-allocation, error-message\
**Created:** [February 23, 2023, 8:34pm UTC](https://discourse.julialang.org/t/outofmemoryerror-in-homotopycontinuation/95105 "2023-02-23T20:34:58Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ricardo\_Miranda\_Mart](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/ricardo_miranda_mart/32/47171_2.png) [@Ricardo\_Miranda\_Mart](https://discourse.julialang.org/u/Ricardo_Miranda_Mart)\
**Post date:** [February 23, 2023, 8:34pm UTC](https://discourse.julialang.org/t/outofmemoryerror-in-homotopycontinuation/95105/1 "2023-02-23T20:34:58Z")

</div>

Hi all,  
I’m trying to use HomotopyContinuation.jl to find all of the solutions a polynomial system of equations with 14 polynomials in 14 variables.  
Each polynomial is sum of monomials of degree from 10 to 13 (like f1=2_x13^2_x14_x15_x16_x24_x25_x26_x34_x35_x36 +  
12_x14_x15_x16_x23_x24_x25_x26_x34_x35_x36+…).

After receiving the error

> **ERROR:** OverflowError: Cannot compute a start system.

I tried to use the option

> start\_system=:total\_degree

in solve, but now I am received the error message

> **ERROR:** OutOfMemoryError()

even without Julia allocating a large amount of memory. What to do in this case?

---

<div class="post-metadata">

**Author:** ![Amine\_o](https://sea2.discourse-cdn.com/julialang/user_avatar/discourse.julialang.org/amine_o/32/209936_2.png) [@Amine\_o](https://discourse.julialang.org/u/Amine_o)\
**Post date:** [December 19, 2024, 1:28am UTC](https://discourse.julialang.org/t/outofmemoryerror-in-homotopycontinuation/95105/2 "2024-12-19T01:28:28Z")

</div>

The start system with total degree homotopy should be z\_j = e^{i2\pi n / p}, for n=0,...,p-1 and for each j. Hence for 14 polynomials in 14 variables that means a starting system with 10^{14} initial starting vectors if p=10. Pretty large given that each path solves a multivariate differential equation of dimension 14. This is highly parallelizable though so it should be possible to deploy machines in parallel to follow each path.