ODE solver return Nan values sometimes

[SergeStinckwich]

This happens for models like the following one :

KendrickModel influenza_england_1978_school
	attribute: #(status -> S I R);
	parameters: #(alpha beta lambda);
	transitions: #( 		S -- lambda --> I. 		I -- alpha --> R. 	);
	lambda: #(beta*I).
	
Composition influenza_england_1978_school_comp
	model:'influenza_england_1978_school'.

Scenario influenza_england_1978_school_scenario
	on: 'influenza_england_1978_school_comp';
	populationSize: 763; S: 738; I: 25;
	alpha:1.46502; beta: 1.232; others: 0.

Simulation influenza_england_1978_school_sim rungeKutta
	scenarios: #(influenza_england_1978_school_scenario);
	from: 3.0; to: 14.0; step: 1; init; execute;
	timeSeriesOutputsAt: '{#status:#I}'

returns a time series with some values Nan.

I guess the problem is related to some bug of in the ODE solver of PolyMath