# Bad accuracy of eigenvalue, how solve it? **URL:** \ **Category:** General Discussion\ **Created:** [May 8, 2024, 3:46pm UTC](https://community.freefem.org/t/bad-accuracy-of-eigenvalue-how-solve-it/3232 "2024-05-08T15:46:20Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![bcaval](https://avatars.discourse-cdn.com/v4/letter/b/d26b3c/32.png) [@bcaval](https://community.freefem.org/u/bcaval)\ **Post date:** [May 8, 2024, 3:46pm UTC](https://community.freefem.org/t/bad-accuracy-of-eigenvalue-how-solve-it/3232/1 "2024-05-08T15:46:20Z")
Dear freefem community, Could you help me to understand my mistake or to advise me **the best way in freefem++ today to make a sure eigenvalue solving ?** Indeed I would like like to make eigenvalue analysis for thermoelasticity and thermopiezoelectricity. I have made the weak forms and now I am testing with the simplest case, well-known…of elasticity. You will find here the \*.msh from GMSH and the \*.edp file. [maillage.7zip](https://community.freefem.org/uploads/short-url/v6sOiBwrAXsUJqjJhI9KlkcA92j.7zip) (368.2 KB) [eigenvalue test.edp](https://community.freefem.org/uploads/short-url/b4MXDZtYedSQjUGQiN6gLGwblO6.edp) (1.6 KB) The frequencies from analytical solution for the flexural modes are: 158 kHz / 993 kHz / 2,78 Mhz / 5,46 MHz with P1 elements it is surestimated by a factor of 1,35! with P2 elements it yields to: 165,8 kHz (~4,9% error)/ 1,03 MHz (~3,7%)/ 2,9 MHz (~4,3%) / … with P3 elements it is similar to P2 case results. **Why such a variation between analytical results and the finite element model ?** But the form of eigenvectors are good (saw on Gmsh postprocessing), the main problem is the accuracy of frequencies! Moreover is the P1 element forbidden for these calculations ? Every advice is welcome. Thanks in advance. Bruno