have actually a Question about thins project? sign uns for a complimentary startupcuba.org accounting to open a issue and also contact itns maintainerns and also the community.

choose a usersurname email address Pasknife authorize uns for startupcuba.org

by click “sign uns because that startupcuba.org”, girlfriend agree come our regards to company and also privacy statement. We’ll sometimes sfinish friend account related emails.

currently ~ above startupcuba.org? authorize in come her account




You are watching: Model failed to converge: degenerate hessian with 1 negative eigenvalues

KennethEnevoldsen commented Dec 1, 2019

one of mine student showed ns the adhering to discrepancy, which i have been able to reproduce:

Ins appears prefer the same design in lme4 and also lmercheck both able come converge (when castle are both singular fit), yet lmercheck it seems to be ~ to throw one error message, which seems choose a bug.

ns lmertest design throwns the error:version failed to converge through 1 negative eigenvalue: -1.6e+03

also though the version obtains the convergence code: 0. I m sorry follow come ns documentatitop top means: convergence code, equal come zero because that successful convergence

here a reduce outPlaced utilizing the recent variation of lme4 and lmertest indigenous CRone and also operation top top MacOns Mojave version 10.14.6:

model making use of lmerTest


hrm2 summary(hrm2)direct combined design fins through preferably likelihood . T-testns use Satterthwaite"ns technique <"lmerModLmerTest">Formula: changeHR_me ~ (HR_self + HR_other) * problem + (problem | participant) + (...)Signif. Codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Correlationship procession not presented by default, together ns = 15 > 12.use print(x, correlation=TRUE) or vcov(x) if you require itconvergence code: 0border (singular) fit: watch ?isSingular">

> hrm2 summary(hrm2)straight combined version fins by preferably likelihood . T-testns use Satterthwaite"ns technique <"lmerModLmerTest">Formula: changeHR_self ~ (HR_self + HR_other) * condition + (problem | participant) + (...)Signif. Codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1Correlationship matrix no displayed by default, together ns = 15 > 12.usage print(x, correlation=TRUE) or vcov(x) if friend need itconvergence code: 0border (singular) fit: see ?isSingular
hrm2_lme4 summary(hrm2_lme4)direct mixed version fit by maximum likelihood <"lmerMod">Formula: changeHR_self ~ (HR_self + HR_other) * problem + (condition | participant) + (...)Correlation procession not shown by default, together ns = 15 > 12.use print(x, correlation=TRUE) or vcov(x) if girlfriend require itconvergence code: 0boundary (singular) fit: check out ?isSingular">

> hrm2_lme4 summary(hrm2_lme4)direct blended model fins by best likelihood <"lmerMod">Formula: changeHR_me ~ (HR_self + HR_other) * condition + (problem | participant) + (...)Correlationship matrix not shown by default, together ns = 15 > 12.usage print(x, correlation=TRUE) or vcov(x) if girlfriend require itconvergence code: 0boundary (singular) fit: watch ?isSingular


*
KennethEnevoldsen discussed thins problem Dec 1, 2019



Owner



KennethEnevoldsen" src="https://avatars.startupcuba.orgusercontent.com/u/23721977?s=88&u=bcdadabce91d4a114a49b811be87c9f952412e89&v=4" />

Copy attach





See more: The Project Type Is Not Supported By This Installation, In Visualstudio 2005